Refactor and improve timeline testing infrastructure

[taldcroft]

Description

This improves the code structure for make_timeline.py and establishes a process for testing.

Highlights:

• Detailed test process with a couple of new utility scripts to help.
• Use dynamic file paths instead of hardwired global paths at import time.
  • In "flight" mode (without --test), input data file paths are hardwired to various paths available on HEAD
  • In "test" mode (with --test), input data files must all be in the directory defined by --data-dir (default t_now).
• Make module importable without side-effects.
• Add a function to grab some files from the web for testing and put them in the --data-dir directory.
• Use CxoTime instead of DateTime.
• Refactor main() into smaller drawing functions. This addresses a longstanding TODO, made easy now with VS code.

Interface impacts

None

Functional testing

On HEAD @ cf7febe

Followed the steps outlined in make_timeline.py for testing on HEAD. This was done with cf7febe. The only changes after that point are documentation or 55cc376 which does not impact the HEAD testing since the --test-get-web option is not used.

After the processing steps, the output timeline states are identical:

$ echo $COMMIT
cf7febe
$ diff t_now/{flight,$COMMIT}/timeline_states.js
$

HTML page outputs:

• https://icxc.cfa.harvard.edu/aspect/test_review_outputs/arc/pr86/t_now/cf7febe/timeline.png
• https://icxc.cfa.harvard.edu/aspect/test_review_outputs/arc/pr86/t_now/flight/timeline.png

Local @ 55cc376

Followed the steps outlined in make_timeline.py for local testing on Mac. This was done with 55cc376. The only changes after that point are documentation or adding two test utility scripts.

After the processing steps, the output timeline states are nearly identical:

$ echo $COMMIT
55cc376
$ diff t_now/{flight,$COMMIT}/timeline_states.yaml    
10c10
< p3_avg_now: '81'
---
> p3_avg_now: '80'
3911c3911
<   fluence: 0.02e9
---
>   fluence: 0.01e9

I'm not 100% sure of the reason for the diff but I suspect that the ACIS fluence page updated in the middle of the test. The change in p3_avg_now (which comes from the ACIS fluence page) would propagate to a small change in fluence which is the predicted fluence assuming constant p3_avg_now flux. There are hundreds of states with a dozen state keys each, and they all match exactly except for this one.

The output web pages and timeline.png for the flight and $COMMIT cases are visually identical.

[jeanconn]

Regarding the new testing plan - I'd previously written up docs in https://github.com/sot/arc/wiki/Set-up-test-Replan-Central for working in a HEAD dev conda environment to test. I'm wondering if maybe make_timeline.py should just have a link to the arc wiki and have a new section for the doing more atomic testing of make_timeline.py or just a link indicating the other docs to test the whole package do exist.

[taldcroft]

@jeanconn - I added a link to the wiki page for full testing of replan central code.

[Copilot]

Copilot reviewed 3 out of 4 changed files in this pull request and generated no suggestions.

Files not reviewed (1)

• pyrightconfig.json: Language not supported