[RFC] Support workflows where the Firehose programmer is already running

[igoropaniuk]

Support two new CLI options to support workflows where the Firehose programmer
is already running on the device:

• --skip-reset (-R): suppresses the post-flash power/reset command, leaving the programmer alive after flashing completes.

• --skip-sahara (-P): skips the Sahara phase entirely and connects directly to an already-running programmer. When used, the programmer ELF argument is omitted. Intended to be paired with --skip-reset from a previous qdl invocation.

Typical use:

# First flash - uploads programmer and flashes, but doesn't reset
qdl prog_firehose_ddr.elf rawprogram*.xml patch*.xml --debug --skip-reset

# Subsequent flashes -  programmer already running, skip Sahara entirely
qdl rawprogram*.xml patch*.xml --skip-sahara

[andersson]

If the device is already in firehose mode, the read in sahara_run() will either return <?xml or a timeout. I think it would be more user friendly to detect these two conditions and from that attempt to continue the firehose session.

The only case where this wouldn't work is if the user interrupted firehose while in rawmode=true, as firehose would read nothing and attempting to write should just end up in flash. I don't know if there's any way around this and that's the reason why I didn't push these patches before.

(As discussed privately) For the --skip-reset I'd prefer such multi-storage flashing to happen in one go if possible - again for better UX.... Perhaps there are other cases where this might be useful? The patch looks good, but I'd prefer that we didn't merge the extra command line arguments until we have that concrete usecase.