Fix qa_Picoscope4000a regressions

[frankosterfeld]

Bump to latest graph-prototype and fix the regressions with Picoscope4000a by adapting to the new lifecycle API.

[frankosterfeld]

This is writing the per-output signal info (SIGNAL_NAME, SIGNAL_UNIT, etc.). This is done manually b/c the block doesn't use the default work()/processBulk() implementation, and instead writes directly to the buffers. Since the latest changes, this only works with the "- 1", leading to an index of -1 instead of 0 when sent with the first data chunk. Is that correct??

[RalphSteinhagen]

-1 is the correct index for the very first tag. Normally you wouldn't need to (and shouldn't) publish this tag explicitly but it's implicitly generated via the property map that is given to the make block routine of the Graph and/or the Block constructor or init(..) functions that take the settings as const property_map &

[frankosterfeld]

We have per-channel settings, and need to do it manually because of the custom work() function, so I think it's needed here. One could argue whether the index should be always -1 (wildcard).

[frankosterfeld]

Shouldn't signal_name and sample_rate automatically get the values from the PS block, via the SIGNAL_NAME/SIGNAL_RATE tag values? (which are received by the monitor, see checks below)

[sonarcloud]

Quality Gate failed

Failed conditions

0.0% Coverage on New Code (required ≥ 80%)

See analysis details on SonarCloud

[frankosterfeld]

Test coverage cannot be improved here, as qa_PicoScope4000a requires the hardware to run...

[wirew0rm]

Looks good, tested locally including the hardware tests.

I also tested the changes with fair-acc/opendigitizer#144, builds without problems. Running it there's still some problems, but I don't think they are likely to be rooted in gr-digitizers.