fix: restore measure filters on comparison accessors (_delta, _delta_perc, _percent_of_total)

[nishantmonu51]

Measure filters on comparison accessors were silently dropped by the unified expression filter refactor (#9746). Restores them.

• The having post-processor puts the raw identifier from the URL (e.g. impressions_delta) into the subquery's measure list. JoinerFilterManager.parse then looked up a measure spec by that suffixed name, found nothing, and dropped the filter.
• Strip the suffix with the existing stripMeasureSuffix helper so the manager is keyed on the base measure. MeasureFilterManager.reconcile already maps the having expression back to the right comparison type via mapExprToMeasureFilter, and commit rebuilds the suffix from the type, so no double suffix and the subquery measure list stays the base measure (same as 0.89).
• Adds tests for _delta, _delta_perc and _percent_of_total: chip created for the base measure with the right type/value, condition applied once, URL param round trips unchanged.

Reproducing steps (on main):

1. Open any explore dashboard with a time comparison enabled (e.g. AdBids, compare to previous period).
2. Add a measure filter: Impressions → % change from → > 10, grouped by Publisher. The URL gets f.AdBids_metrics=publisher having (impressions_delta_perc GT 0.1).
3. Reload the page, or paste that URL directly.
4. Before: no measure chip is shown and the leaderboards are unfiltered. After: the Impressions chip shows % change > 10 and the filter is applied.

The same happens for impressions_delta gt 10 (absolute change) and impressions_percent_of_total gt 0.25 (percent of total). Plain impressions gt 10 was unaffected.

Checklist:

• Covered by tests
• Ran it and it works as intended
• Reviewed the diff before requesting a review
• Checked for unhandled edge cases
• Linked the issues it closes
• Checked if the docs need to be updated. If so, create a separate Linear DOCS issue
• Intend to cherry-pick into the release branch
• I'm proud of this work!