Add support for `SHOW TOTALS` metrics in the PgBouncer input plugin

### Use Case

I use the Telegraf PgBouncer plugin to collect connection pool statistics. In environments with many pools/databases configured in PgBouncer, it becomes difficult to manage and analyze the metrics per-database. Having aggregated metrics would help monitor overall PgBouncer performance more efficiently and reduce metric cardinality in the output backend.

PgBouncer provides such aggregated stats via the `SHOW TOTALS` command. These totals are very useful, especially in large-scale setups.

### Expected behavior

Telegraf should collect PgBouncer totals (aggregated stats across all pools) by executing `SHOW TOTALS` and expose them in a new measurement, e.g., pgbouncer_totals. The fields would be similar to the existing pgbouncer measurement, but without the db tag.

Alternatively, this could be implemented by keeping the same pgbouncer measurement but using a special tag such as `db=__totals__ `or `db=TOTALS`.

### Actual behavior

Currently, the Telegraf PgBouncer plugin only collects per-database statistics using `SHOW STATS`. There is no way to collect aggregated totals directly. Computing global values manually is not accurate, especially for `avg_*_time` fields, which cannot be averaged across databases meaningfully.

### Additional info

PgBouncer reference: https://www.pgbouncer.org/usage.html#show-totals
Telegraf PgBouncer plugin: https://github.com/influxdata/telegraf/blob/master/plugins/inputs/pgbouncer/README.md

Start by inspecting the PgBouncer input plugin and its documentation in plugins/inputs/pgbouncer/README.md, then trace the existing SHOW STATS collection path. Define whether totals use a new measurement or a special db tag, implement collection of SHOW TOTALS fields, and verify that aggregated metrics omit or distinguish the database tag.

Written by the indexing model from the issue text.