Stabilize STLink probe source

[tmplt]

As mentioned in f2a73da, the new CMSIS-DAP implementation is buggy. Investigate the cause and fix.

[tmplt]

Tangent: an STLink is not a CMSIS-DAP. DAPSource should be further abstracted to ProbeSource. We do not care what kind of probe is ultimately used, only that probe-rs supports it.

[tmplt]

Mostly we just get an invalid sync size. This could just be a variance from the ITM standard.

[tmplt]

Perhaps a feature-set of quirks should be maintained upstream?

[tmplt]

This would be an upstream issue in probe-rs. Presumably the amount of work is non-trivial. Postponing.

[tmplt]

I theorize that the root cause of this issue is transient configuration noise seen on the SWO pin. So far I've only been able to reproduce this issue on a Nucleo, where the SWO pin feeds into an on-chip debugger that is interfaced with via USB. Verifying if this indeed is the cause should be trivial, but I do not have a Nucleo at hand at the moment.