feat(CountBasedMetadata): add notes to example signal panels

[saiflayouni]

Summary

Closes #235

The HTML report shows present_examples and total_examples diagnostics pointing to the same locations, and the total_examples metadata count (e.g. 37) differs from the number of paths shown in the diagnostics (e.g. 27). This is correct engine behavior — diagnostics mark unique locations, while the metadata counts individual example objects — but was opaque to users.

• Adds an optional note field to CountConfig in CountBasedMetadata.tsx
• Renders the note in both the donut-chart branch (response_coverage) and the progress-bar branch (all other count signals), replacing the previous hardcoded string
• For example_density: explains that expected_examples and present_examples diagnostics list overlapping location sets (they converge when all expected locations already have examples)
• For example_validity: explains that the total-examples count in the summary can exceed the diagnostic location count when a single location holds multiple examples

Test plan

• All 75 existing @jentic/api-scorecard-formatter-html unit tests pass (npm test -w @jentic/api-scorecard-formatter-html)
• signals.test.tsx SSR tests for example_density and example_validity still pass (no undefined/NaN, expected numerator/denominator values)
• response_coverage note text is unchanged (moved from hardcoded JSX to COUNT_SIGNAL_CONFIG)

[saiflayouni]

Thanks for the detailed explanation in #235, @char0n! The example_density note I had originally written implied expected_examples and present_examples overlap, which was wrong. Corrected in 6d3fda8 — it now describes them as superset/subset: expected_examples marks every element that can hold a value; present_examples is the subset that already does.

The example_validity note (total counts values, paths count elements) matches your explanation and is unchanged.

I saw #245 is open for the broader tooltip/help-text work — happy to help there too once #42 lands definitions.