Add labels, tooltips and verdicts for beat error, rate error, amplitude and beat rate

[xelrix123]

This is a PR to implement changes described in issue #13 .

NOTE: Claude code was used to implement rough changes (it's been a while I've touched C codebase, so....) and then few things were manually tweaked. However I am not familiar with specifics of GTK3 etc. UI code so even though tooltip and label additions look ok to me (and only a bit oberly verbose) -- caveat emptor. All the changes/updates are to src/output_panel.c .

Proposed verdict ranges are:

rate error:
        if (abs_rate <= 5.0) return "excellent";
        if (abs_rate <= 10.0) return "good";
        if (abs_rate <= 20.0) return "acceptable";
        if (abs_rate <= 30.0) return "marginal";
        return "poor";

and

beat error:
        if (be_ms <= 0.5) return "excellent";
        if (be_ms <= 1.0) return "good";
        if (be_ms <= 2.0) return "acceptable";
        if (be_ms <= 3.0) return "poor";
        if (be_ms <= 5.0) return "bad";
        return "severe";

Some notes:

• these changes were implemented in a way that current UI layout wasn't touched: as a result fonts for labels and verdicts are somewhat small and especially on very small or very large displays could become illegible (tooltips mitigate this somewhat). In a future, a slightly higher panel would make the labels / verdicts a bit more readable.
• tooltip contents should be reviewed for correctness (as not to accidentally provide incorrect advice)
• tooltips have logic flaw of ranges overlapping (e.g. 0-1,1-2 etc.). But in code this is not so (see above) and was done to keep tooltips simpler.
• verdicts: are these good ranges? I am in no way attached to the ones in this PR and they are based merely on searching online, and double checking with the LLMs to provide something that would be helpful. I warmly welcome more experienced opinions on useful ranges / verdicts to display (see also discussion in issue UI: add tooltips. labels and verdicts for rate error, beat error, amplitude and beat rate #13 about whether having a verdict could in some cases backfire as a completely incorrect interpretation of what is reported).

Testing:

• Linux: Pop!_OS 24.04 LTS / 13" laptop @ 1080p: compiles / runs, labels are legible
• Windows 11 (using Mingw) / wrokstation @ 4k display: compiles / runs, however the fonts in the panel approach illegible
• OSX: not tested
• Live view: tooltips, verdicts work
• Snapshots: tooltips work verdicts are frozen and are shown

---

As a personal note - I am using tg at complete random intervals and have at best cursory understanding of watch movement mechanics and things that can be measured with timegrapher. Thus it is quite possible that I have completely missed some important considerations why this kind of feature doesn't really fit the tg-timer. In such a case, don't hesitate to point out errors in my thinking. And of course in the code...

Cheers!