[Beforerr]

One commit per review item; each reviews on its own (Commits tab).

Commit	What
remove unit	threw on any multi-character string; no user in the org
broadcast shape	DataVariable(rand(3)) .+ ones(3, 2) threw DimensionMismatch
localize age	compared local now() against UTC mtime; now mtime + age < now(UTC), so Month works too
==/hash	unique kept equal FilePatterns; Dataset == threw across type params; unused ValidityChecks isequal/hash dropped
listings	404/410 (curl 37 for file://) → cached empty listing, silent; other failures warn and are evicted so the next call retries
DOY render	FilePattern(...)("2020-275") now parses like remotefiles
parse_datetime	space or / separator via same-shape DateFormats; +3–9 ns
units via schema	ISTP UNITS found; DefaultSchema :unit → "units" (CF). Plot labels in SpacePhysicsMakie pick up "units"
README	drop Product lines (default reader returns paths)
refactor	_resolve_steps shared by remotefiles/available; _format(::String, fills) → _fill_keys; dead ismissing
cadence	ArgumentError instead of @assert
SchemaLookup haskey	agrees with get

Tests: 280/280 locally. SpaceWeather's GOES registry (Int defaults) selects as before, and filter(SGPS; id=(16, 18)) works.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[github-actions]

Benchmark Results (Julia v1)

Time benchmarks

	main	29590db...	main / 29590db...
mask/fill in range	0.0496 ± 0.00075 ms	0.0494 ± 0.00057 ms	1 ± 0.019
mask/one value per check	30.2 ± 0.4 μs	30.1 ± 0.3 μs	1 ± 0.017
mask/per component	0.0414 ± 0.00073 ms	0.041 ± 0.00078 ms	1.01 ± 0.026
time_to_load	0.106 ± 0.0025 s	0.105 ± 0.0014 s	1.01 ± 0.027

Memory benchmarks

	main	29590db...	main / 29590db...
mask/fill in range	11 allocs: 1.09 kB	11 allocs: 1.09 kB	1
mask/one value per check	5 allocs: 0.188 kB	5 allocs: 0.188 kB	1
mask/per component	9 allocs: 0.781 kB	9 allocs: 0.781 kB	1
time_to_load	0.2 k allocs: 11.7 kB	0.2 k allocs: 11.7 kB	1