nrf51.cpu -- clearing lockup after double fault

Running `apache-mynewt-nimble/apps/bleprph` app on a micro:bit v1 yields the following problem:

```
$ newt run myperiph 0
App image successfully generated: ...
Loading app image into slot 1 ...
; nrf51.cpu configure -event gdb-detach {if {[nrf51.cpu curstate] eq "halted"} resume;shutdown}
GNU gdb (GNU Arm Embedded Toolchain 10.3-2021.10) 10.2.90.20210621-git

--Type for more, q to quit, c to continue without paging--

For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from ...
0xfffffffe in ?? ()
target halted due to debug-request, current mode: Thread
xPSR: 0xc1000000 pc: 0xfffffffe msp: 0xfffffffc
0xfffffffe in ?? ()
target halted due to debug-request, current mode: Thread
xPSR: 0xc1000000 pc: 0xfffffffe msp: 0xfffffffc
(gdb) c
Continuing.
nrf51.cpu -- clearing lockup after double fault

Program received signal SIGINT, Interrupt.
0xfffffffe in ?? ()
```

Setup:

```
# target.yml
target.app: "@apache-mynewt-nimble/apps/bleprph"
target.bsp: "@apache-mynewt-core/hw/bsp/bbc_microbit"
target.build_profile: optimized
```
```
# syscfg.yml
syscfg.vals:
LOG_LEVEL: 255
BLE_LL_CFG_FEAT_LE_ENCRYPTION: 0
BLE_SM_LEGACY: 0
BLE_SM_SC: 0
```

```
# project.yml

...

repository.apache-mynewt-core:
type: github
vers: 0-dev
user: apache
repo: mynewt-core
```

The same problem occurs with other example apps.

Reproduce with apache-mynewt-nimble/apps/bleprph using the target.yml, syscfg.yml, and project.yml settings shown, on the bbc_microbit BSP. Start by tracing the nrf51.cpu/GDB halt output and compare it with other example apps; done means identifying the cause of the double-fault lockup and verifying a reliable recovery on the micro:bit v1.

Written by the indexing model from the issue text.