Retrying commands that result in a FAULT flag being set is problematic for the NXP IMXRT685s:

When issuing SYSRESETREQ to AIRCR the target responds with a FAULT flag @ https://github.com/probe-rs/probe-rs/blob/master/probe-rs/src/vendor/nxp/sequences/nxp_armv8m.rs#L687-L696. Probe-rs retries the command for 5 times in this case. https://github.com/probe-rs/probe-rs/blob/master/probe-rs/src/probe/cmsisdap/mod.rs#L523

Depending on the speed of the host machine, this presumably results the target being reset in the bootloader, which causes the target to no longer respond (hence the WAIT states). Validated this by temporarily removing retries on a slow host, removing the issue, and introducing the issue by adding a 1ms delay after a FAULT in the batch processor.

Basically, on my slower laptop, probe-rs does not work at all when programming/debugging the NXP IMXRT685s.

With this PR I want to get the conversation started:

• do we even want to retry commands that result in FAULT flags?
• if no, will removing the retry cause probe-rs to be less reliable?
• if the answer is "it depends", how do we want to change the API such that we can either configure the retry count, or disable retries for specific parts in a sequence.

Note that in the polyfill probe only WAIT, not FAULT, is retried. So if we want it in Cmsisdap probe, then probably we want it in the polyfill probe too.

Original issue: #3248 (comment)