fix two performance bottlenecks in the gribjump source

[andreas-grafberger]

Fixes two small performance bottlenecks that made retrieval for large time-series unnecessarily slow:

1. If fetch_coords_from_fdb=True, the reference latitudes and longitudes were read from the reference field again for each individually retrieved field. Originally, I somehow assumed they would be cached together with the GribMetadata object and only noticed this issue recently during some benchmarks. I propose to simply fix this for now by explicitly caching latitudes/longitudes.
2. If indices instead of ranges were used, they would get converted into a list of ranges (by this function) again for each single field. As a simple fix, I propose that we just convert indices to ranges once directly in this source. I will follow up on this in a future PR to find a cleaner fix.

I performed a few ad-hoc benchmarks for these two fixes:

1. Retrieving one month of hydrological reanalysis data shows that the wall-clock time with the same configuration (same fdb, #threads, ~1k random points) goes from ~1 min to ~18 seconds.
2. When retrieving 4 years of hydrological reanalysis data, ~14% of that time were spent in repeated calls to (ExtractionRequest.from_indices). This fix would remove this overhead.

Contributor Declaration

By opening this pull request, I affirm the following:

• All authors agree to the Contributor License Agreement.
• The code follows the project's coding standards.
• I have performed self-review and added comments where needed.
• I have added or updated tests to verify that my changes are effective and functional.
• I have run all existing tests and confirmed they pass.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.35%. Comparing base (4acd81b) to head (b44dce5).

Additional details and impacted files

@@           Coverage Diff            @@
##           develop     #877   +/-   ##
========================================
  Coverage    85.35%   85.35%           
========================================
  Files          184      184           
  Lines        14562    14562           
  Branches       732      732           
========================================
  Hits         12429    12429           
  Misses        1922     1922           
  Partials       211      211           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[ChrisspyB]

seems to make sense

[sandorkertesz]

@andreas-grafberger thank you for this improvement. Please synch this PR with develop and it can be merged.

[andreas-grafberger]

@andreas-grafberger thank you for this improvement. Please synch this PR with develop and it can be merged.

@sandorkertesz Thank you! I rebased onto develop and the gribjump tests run locally on my machine with the latest commit.

From my side, either merging or waiting for #878 would be fine, which enables the gribjump tests in the CI. There are still a few hurdles I ran into, which will delay this by a couple of days.