[R] Flaky behavior with multiple to_duckdb() calls in one pipeline

### Describe the bug, including details regarding any error messages, version, and platform.

Hello, it's me again :)

Not with a big-data crash today, but a different oddity.
I have a large-ish parquet dataset, which I process with multiple dplyr verbs, including `inner_join`s.
I used to do this in "plain" Arrow, but wanted to try `to_duckdb` for performance (and also because it's cool).
I join the dataset to a list of characters and later to a tibble.
`join_by` complains if the backends are not the same (I could provide `copy=TRUE`), so I decided to move those to DuckDB, too.

However, I noticed that when I execute multiple queries with `inner_join(foo %>% to_duckdb(), ...)`s,
DuckDB sometimes doesn't find the referenced table. It's flaky.

This is the error:
```R
Error in `collect()`:
! Failed to collect lazy table.
Caused by error in `dbSendQuery()`:
! rapi_prepare: Failed to prepare query SELECT reason, SUM(n) AS n
FROM (
SELECT LHS.*, reason
FROM (
SELECT cid, COUNT(*) AS n
FROM (
SELECT arrow_107.*
FROM arrow_107
INNER JOIN arrow_108
ON (arrow_107.cid = arrow_108.cid)
) q01
GROUP BY cid
) LHS
LEFT JOIN arrow_109
ON (LHS.cid = arrow_109.cid)
) q01
GROUP BY reason
Error: Catalog Error: Table with name arrow_108 does not exist!
Did you mean "pg_proc"?

LINE 9: INNER JOIN arrow_108
^
Run `rlang::last_trace()` to see where the error occurred.
> rlang::last_trace()

Error in `collect()`:
! Failed to collect lazy table.
Caused by error in `dbSendQuery()`:
! rapi_prepare: Failed to prepare query SELECT reason, SUM(n) AS n
FROM (
SELECT LHS.*, reason
FROM (
SELECT cid, COUNT(*) AS n
FROM (
SELECT arrow_107.*
FROM arrow_107
INNER JOIN arrow_108
ON (arrow_107.cid = arrow_108.cid)
) q01
GROUP BY cid
) LHS
LEFT JOIN arrow_109
ON (LHS.cid = arrow_109.cid)
) q01
GROUP BY reason
Error: Catalog Error: Table with name arrow_108 does not exist!
Did you mean "pg_proc"?

LINE 9: INNER JOIN arrow_108
^
---
Backtrace:
▆
1. ├─... %>% collect()
2. ├─dplyr::collect(.)
3. └─dbplyr:::collect.tbl_sql(.)
4. ├─base::withCallingHandlers(...)
5. ├─dbplyr::db_collect(...)
6. └─dbplyr:::db_collect.DBIConnection(...)
7. ├─DBI::dbSendQuery(con, sql)
8. └─duckdb::dbSendQuery(con, sql)
```

The particularly weird thing is that it usually works the first time,
but fails when I execute it multiple times one after the other.
Is there some GC mechanism at work maybe that kicks in during preparation/execution of the second run?

Reproducer:
```R

# Probably unrelated, but this is how Arrow was compiled:
Sys.setenv("ARROW_WITH_ZLIB"="ON")
Sys.setenv("ARROW_PARQUET" = "ON")
Sys.setenv("ARROW_WITH_ZSTD" = "ON")
Sys.setenv("ARROW_MIMALLOC" = "ON")
Sys.setenv("LIBARROW_MINIMAL" = FALSE)
Sys.setenv("LIBARROW_BINARY" = FALSE)
Sys.setenv("ARROW_R_DEV" = TRUE)
Sys.setenv(MAKEFLAGS = sprintf("-j%d", parallel::detectCores()))
options(renv.config.pak.enabled = TRUE)

# Dependencies:
library(arrow)
library(dplyr)
# I might have forgotten a package here, this is part of a larger project

foo1 <- arrow_table(cid = c("a","b","c", "d", "a"), peer = c("1","2","3","2","2"))
foo2 <- c("b","c")
foo3 <- tibble(cid=c("a","b","c","d"), reason=c("something","something else","something else","something"))

# This fails most of the time.
# In particular if you execute it multiple times quickly in succession.
foo1 %>%
to_duckdb() %>%
inner_join(arrow_table(tibble(cid=foo2)) %>% to_duckdb(), by=join_by(cid)) %>%
group_by(cid) %>%
tally() %>%
left_join(arrow_table(foo3) %>% to_duckdb(), by=join_by(cid)) %>%
group_by(reason) %>%
summarize(n=sum(n)) %>%
collect()

# This works more often, but still fails sometimes:
foo1 %>%
to_duckdb() %>%
inner_join(arrow_table(tibble(cid=foo2)) %>% to_duckdb(), by=join_by(cid)) %>%
group_by(cid) %>%
tally() %>%
left_join(arrow_table(foo3) %>% to_duckdb(), by=join_by(cid)) %>%
collect()

# Pulling out the inline to_duckdb calls makes it so that it seemingly always works:
tmp1 <- arrow_table(tibble(cid=foo2)) %>% to_duckdb()
tmp2 <- arrow_table(foo3) %>% to_duckdb()

foo1 %>%
to_duckdb() %>%
inner_join(tmp1, by=join_by(cid)) %>%
group_by(cid) %>%
tally() %>%
left_join(tmp2, by=join_by(cid)) %>%
group_by(reason) %>%
summarize(n=sum(n)) %>%
collect()
```

### Additional Info

#### Machine Overview:
```
Memory: 378 GB
CPU: 64x Intel(R) Xeon(R) Gold 6154
OS: Debian 12
```

#### R `sessionInfo()`
```R
> sessionInfo()
R version 4.5.0 (2025-04-11)
Platform: x86_64-pc-linux-gnu
Running under: Debian GNU/Linux 12 (bookworm)

Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.11.0
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.11.0 LAPACK version 3.11.0

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C LC_TIME=en_US.UTF-8
[4] LC_COLLATE=en_US.UTF-8 LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C LC_ADDRESS=C
[10] LC_TELEPHONE=C LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

time zone: Europe/Berlin
tzcode source: system (glibc)

attached base packages:
[1] stats graphics grDevices datasets utils methods base

other attached packages:
[1] treemapify_2.5.6 paletteer_1.6.0 ggplot2_3.5.2 viridis_0.6.5
[5] viridisLite_0.4.2 pracma_2.4.4 xtable_1.8-4 forcats_1.0.0
[9] readr_2.1.5 arrow_20.0.0 tidyr_1.3.1 stringr_1.5.1
[13] lubridate_1.9.4 dplyr_1.1.4

loaded via a namespace (and not attached):
[1] utf8_1.2.6 generics_0.1.4 renv_1.0.3 stringi_1.8.7
[5] hms_1.1.3 magrittr_2.0.3 grid_4.5.0 timechange_0.3.0
[9] RColorBrewer_1.1-3 blob_1.2.4 DBI_1.2.3 rematch2_2.1.2
[13] gridExtra_2.3 purrr_1.0.4 scales_1.4.0 duckdb_1.3.0
[17] cli_3.6.5 rlang_1.1.6 crayon_1.5.3 dbplyr_2.5.0
[21] bit64_4.6.0-1 withr_3.0.2 tools_4.5.0 parallel_4.5.0
[25] tzdb_0.5.0 assertthat_0.2.1 vctrs_0.6.5 R6_2.6.1
[29] lifecycle_1.0.4 bit_4.6.0 vroom_1.6.5 pkgconfig_2.0.3
[33] pillar_1.10.2 gtable_0.3.6 glue_1.8.0 ggfittext_0.10.2
[37] tibble_3.3.0 tidyselect_1.2.1 rstudioapi_0.17.1 farver_2.1.2
[41] compiler_4.5.0
```

#### `lsb_release -a`
```
No LSB modules are available.
Distributor ID: Debian
Description: Debian GNU/Linux 12 (bookworm)
Release: 12
Codename: bookworm
```

#### `uname -a`
```
Linux 6.1.0-34-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.135-1 (2025-04-25) x86_64 GNU/Linux
```

#### `cat /proc/cpuinfo` (truncated)
```
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 85
model name : Intel(R) Xeon(R) Gold 6154 CPU @ 3.00GHz
stepping : 4
microcode : 0x2007108
cpu MHz : 2992.968
cache size : 16384 KB
physical id : 0
siblings : 64
core id : 0
cpu cores : 64
apicid : 0
initial apicid : 0
fpu : yes
fpu_exception : yes
cpuid level : 22
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon rep_good nopl xtopology cpuid tsc_known_freq pni pclmulqdq vmx ssse3 fma cx16 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single pti ssbd ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm mpx avx512f avx512dq rdseed adx smap clflushopt clwb avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves arat umip pku ospke md_clear flush_l1d arch_capabilities
vmx flags : vnmi preemption_timer posted_intr invvpid ept_x_only ept_ad ept_1gb flexpriority apicv tsc_offset vtpr mtf vapic ept vpid unrestricted_guest vapic_reg vid shadow_vmcs pml tsc_scaling
bugs : cpu_meltdown spectre_v1 spectre_v2 spec_store_bypass l1tf mds swapgs taa mmio_stale_data retbleed gds bhi ibpb_no_ret
bogomips : 5985.93
clflush size : 64
cache_alignment : 64
address sizes : 46 bits physical, 48 bits virtual
power management:
```

### Component(s)

R