CI: Cache the earth_day_01d_p grid

[seisman]

Description of proposed changes

The earth_day_01d_p.tif file is used in test_grdimage_image.py.

[seisman]

One quick thought when working on this PR.

Currently, the "Cache Data" workflow is disabled by default for PRs. Sometimes, when we add a new cache file, we need to manually enable the workflow in PRs to refresh the caches and also need to disable it before merging. It's tedious although it doesn't happen very often.

A better solution is:

1. Move the download_test_data function to a separate source file, e.g., pygmt/helpers/caching.py.
2. Enable the workflow if pygmt/helpers/caching.py changes:

pull_request:
paths:
- 'pygmt/helpers/caching.py'

[codspeed-hq]

CodSpeed Performance Report

Merging #2937 will degrade performances by 6.45%

_{⚠️ No base runs were found}

_{Falling back to comparing cache/earth_day (2ceb6eb) with main (29f87b7)}

Summary

❌ 1 regressions
✅ 63 untouched benchmarks

⚠️Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	cache/earth_day	Change
❌	test_grdlandmask_no_outgrid	2.3 s	2.4 s	-6.45%

[seisman]

CodSpeed Performance Report

Merging #2937 will degrade performances by 6.45%

⚠️ No base runs were found

Falling back to comparing cache/earth_day (2ceb6eb) with main (29f87b7)

Summary

❌ 1 regressions ✅ 63 untouched benchmarks

⚠️Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

Benchmark maincache/earth_day Change
❌ test_grdlandmask_no_outgrid 2.3 s 2.4 s -6.45%

FYI, I've changed the regression threshold back to 10% to silence false warnings like the above:

[weiji14]

FYI, I've changed the regression threshold back to 10% to silence false warnings like the above:

Thanks, I've opened an issue (#2942) to track why this benchmark test seems to have so many false warnings.