Time stepped ground filter

[keskitalo]

Time-stepped ground filter and all the facilities to perform map co-add in memory.

[Copilot]

Pull request overview

This PR adds time-stepped ground filtering and expands infrastructure for in-memory HEALPix map co-addition / comparison, while also standardizing operator timing logs across multiple TOAST ops.

Changes:

• Extend FilterBin with time-stepped ground templates, template covariance handling controls, and facilities for saving template examples / amplitudes.
• Refactor toast_healpix_coadd.py to support cached (in-memory) inputs, optional RING output, and more flexible weighting.
• Add a new toast_healpix_compare.py script and broaden operator-level “Applying/Applied” timing logs.

Reviewed changes

Copilot reviewed 21 out of 21 changed files in this pull request and generated 39 comments.

Show a summary per file

File	Description
src/toast/tests/ops_filterbin.py	Updates FilterBin test configuration to exercise new options.
src/toast/scripts/toast_healpix_compare.py	New CLI script to compare two HEALPix maps.
src/toast/scripts/toast_healpix_coadd.py	Refactor coadd workflow (cache/result support, ring output option, input parsing changes).
src/toast/scripts/CMakeLists.txt	Installs the new toast_healpix_compare.py entrypoint.
src/toast/pixels_io_healpix.py	Simplifies HDF5 read argument handling and removes verbose printouts.
src/toast/ops/time_constant.py	Adds consistent operator start/end timing logs.
src/toast/ops/simple_statcut.py	Adds consistent operator start/end timing logs and adjusts logging output.
src/toast/ops/simple_jumpcorrect.py	Adds consistent operator start/end timing logs.
src/toast/ops/simple_deglitch.py	Adds consistent operator start/end timing logs and adjusts logging output.
src/toast/ops/signal_diff_noise_model.py	Adds consistent operator timing logs.
src/toast/ops/scan_healpix.py	Adds consistent operator timing logs.
src/toast/ops/polyfilter/polyfilter.py	Adds consistent operator timing logs.
src/toast/ops/noise_model.py	Adds consistent operator timing logs in FlagNoiseFit.
src/toast/ops/mapmaker.py	Changes initialization of START/STOP header aggregation values.
src/toast/ops/hwpfilter.py	Adds reverse mode and consistent operator timing logs; refactors template subtraction path.
src/toast/ops/filterbin.py	Implements time-stepped ground templates, template metadata capture, rcond-based handling, template saving, and output naming changes.
src/toast/ops/demodulation.py	Fixes help text typo and adds consistent operator timing logs.
src/toast/ops/decimate_focalplane.py	New operator to decimate (flag) detectors on the focalplane.
src/toast/ops/calibrate.py	Adds consistent operator timing logs.
src/toast/ops/init.py	Exports the new DecimateFocalplane operator.
src/toast/ops/CMakeLists.txt	Installs the new decimate_focalplane.py module.

---

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

[Copilot]

@keskitalo I've opened a new pull request, #917, to work on those changes. Once the pull request is ready, I'll request review from you.

[Copilot]

@keskitalo I've opened a new pull request, #918, to work on those changes. Once the pull request is ready, I'll request review from you.

[Copilot]

@keskitalo I've opened a new pull request, #921, to work on those changes. Once the pull request is ready, I'll request review from you.

[Copilot]

@keskitalo I've opened a new pull request, #922, to work on those changes. Once the pull request is ready, I'll request review from you.

[keskitalo]

There are intermittent unit test failures but they seem unrelated to the PR.

[tskisner]

All three test failures happen in the same place in the ops_filterbin.py test file. Do they pass locally? Do they pass on the toast3 branch?

[tskisner]

I'll just note that this test fails with the same error on my local system:

OMP_NUM_THREADS=1 mpirun -np 2 python -c 'import toast.tests; toast.tests.run("ops_filterbin")'

<snip>

Proc 0: Traceback (most recent call last):
Proc 0:   File "/home/kisner/software/conda/envs/toast/lib/python3.12/unittest/case.py", line 58, in testPartExecutor
    yield
Proc 0:   File "/home/kisner/software/conda/envs/toast/lib/python3.12/unittest/case.py", line 634, in run
    self._callTestMethod(testMethod)
Proc 0:   File "/home/kisner/software/conda/envs/toast/lib/python3.12/unittest/case.py", line 589, in _callTestMethod
    if method() is not None:
       ^^^^^^^^
Proc 0:   File "/home/kisner/software/conda/envs/toast/lib/python3.12/site-packages/toast/tests/ops_filterbin.py", line 776, in test_filterbin_obsmatrix_noiseweighted
    fname_matrix = ops.coadd_observation_matrix(
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Proc 0:   File "/home/kisner/software/conda/envs/toast/lib/python3.12/site-packages/toast/ops/obsmat.py", line 187, in coadd_observation_matrix
    raise RuntimeError(msg)
Proc 0: RuntimeError: Cannot find an inverse covariance matrix to go with '{infile_matrix}'

[0]error Abort(1) on node 0 (rank 0 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0

The same command running on an install of the toast3 branch works, so the cause of the failure is something in this PR. I'll see if anything stands out as I look through it.