misc. bugfixes

[meeg]

Some quick fixes that have piled up:

• Firmware Updates #367 had a typo which led to phase coherence being lost on generators with mixers - the generator phase would be coherent within the execution of a single program, but would change randomly with each new program even if the mixer_freq was kept the same. Thanks to various users for reporting this and Jeffrey Birenbaum of MIT-LL for identifying the cause.
• When the reps_innermost option for AveragerProgramV2 was used, the acquired data was being averaged over the wrong axis. Thanks to @probvar for catching this and proposing a fix.
• changes to improve multiboard sync #342 broke the tProc v1 acquisition code for raw (shot-by-shot) data - the raw data arrays would have an unexpected shape. Fixed. Thanks to Trevyn Larson of NIST for reporting this.
• The AxisSignalGenV6Ctrl driver got broken by some change. This driver is only used in a special firmware used by the QICK core team, so this should not be relevant for anyone else.