emscope should output JSON

[gcohler]

• output in JSON format -- with SI units wherever possible -- e.g. watts not uWatts
• include more in the JSON -- so that the output stands on its own. See example below.
• avoid abbreviations that are hard to understand
• use real hardware part numbers and software part numbers -- so that we know exactly what was tested in case new versions come out.

example:

{
    "timestamp": 1755509250,
    "measurement_device": "Joulescope JS220",
    "unit_under_test": {
        "manufacturer": "Texas Instruments",
        "board_tested": "LP-EM-CC2340R5",
        "software_tested": "zephyr v10.0",
        "voltage_supplied": 3.3
    },
    "average_sleep_watts":   0.000001941,
    "event_rate_hertz": 1,
    "joules_per_representative_event": 0.000030980,
    "joules_per_cycle": 0.000030921,
    "joules_per_day": 2.844,
    "EMeralds": 28.13
}

[biosbob]

to clarify, these files are internal artifacts.... i originally imagined they would be augmented by a user, but backed away.... having them in yaml just makes them more readable (for me)....

the capture.yaml file is absolutely minimal (by design); i might add a SHA for integrity.... more important, this file is only written (once) by emscope grab....

the analysis.yaml is likewise minimal -- though it can be updated implicitly through emscope scan.... the file even records the set of filters applied when this file was created.... a (unfiltered) version of this is created through emscope grab.... subsequent calls to emscope scan would "refine" this file applying a variety of filters -- trimming, coalescing events, finer-grained event demarcation, etc....

[gcohler]

I'm not referring to the yaml files. Yaml is fine vs json. I'm talking about stdout outputs.