Fix fill_bad_time_intervals

[matteobachetti]

As reported at Lorentz Center, this function had a few problems When multiple short BTIs were present, it often failed to measure a robust count rate. It did not reject events from the bad intervals before filling them, adding events to the existing ones and overestimating the number of counts. Also, it was very slow in problematic NICER datasets.

The new function:

1. Eliminates sequences of very short GTI/BTIs, as those intervals are problematic anyway.
2. Eliminates the existing counts from BTIs, avoiding duplication.
   The decrease of the number of BTIs is sufficient to greatly speed up the computation in the Spectral Timing tutorial.
   The result is below:

The green shows the filled event list, the black is the raw one. Notice how the intervals that were incorrectly filled in the old version (e.g. the large drop at time ~3320) are now marked as BTIs (a feature, not a bug; simply, a very bad time interval that should not be trusted). Also, when filled, the light curve has a slightly lower value than before, closer to the average count rate that should be considered the stable flux of the source.

[Copilot]

Pull request overview

This PR fixes multiple issues with the fill_bad_time_intervals function that were reported at the Lorentz Center. The main problems addressed are:

1. Failure to measure robust count rates when multiple short Bad Time Intervals (BTIs) were present
2. Incorrect addition of events from bad intervals to existing ones, causing overestimation
3. Very slow performance on problematic NICER datasets

Changes:

• Added new eliminate_short_gtis function to remove sequences of very short GTIs that are problematic
• Modified fill_bad_time_intervals to eliminate existing counts from BTIs before filling, and to skip sequences of short GTIs
• Improved analyze_segments to use consistent even-sampling detection logic

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.

File	Description
stingray/gti.py	Added eliminate_short_gtis function to filter out problematic short GTI sequences
stingray/base.py	Fixed fill_bad_time_intervals to properly handle BTIs, updated buffer size calculation, added GTI filtering, sorted random events, changed BTI visualization, and improved analyze_segments even-sampling detection
stingray/tests/test_base.py	Updated tests to use explicit max_length and buffer_size parameters
docs/changes/968.bugfix.rst	Added changelog entry for the bug fix

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.18%. Comparing base (79b8f9b) to head (9b56019).
⚠️ Report is 9 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #968      +/-   ##
==========================================
- Coverage   96.21%   96.18%   -0.03%     
==========================================
  Files          48       48              
  Lines        9949     9987      +38     
==========================================
+ Hits         9572     9606      +34     
- Misses        377      381       +4     

☔ 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.

[matteobachetti]

Changes included here for inspection: StingraySoftware/notebooks#127

[eleonorav89]

No comment on my side. Thanks for your work!