mmt_binospec: split setups by mask, require INCAN=on for flats

[tepickering]

Fixes #2137.

Problem 1 — different slit masks in one setup

mmt_binospec.configuration_keys() only returned ['dispname'], so all frames with the same grating were grouped into one setup regardless of slit mask, mixing slit traces, pixel flats, and wavelength solutions across mask geometries.

Fix: add decker (FITS MASK) to configuration_keys() so different masks get separate setups and calibration groups. Also propagate MASK via raw_header_cards().

Problem 2 — INCAN=off frames typed as flats

check_frame_type() for pixelflat/trace/illumflat only required HENEAR=off and SCRN=deployed, so dark/noisy INCAN=off exposures were auto-typed as flats.

Fix: map the incandescent lamp as lampstat03 (FITS INCAN) in init_meta() and require INCAN=on in the flat/trace branch of check_frame_type().

Tests

New pypeit/tests/test_binospec_setup.py covering:

• decker is in configuration_keys()
• INCAN=on (screen deployed, arc off) → typed as pixelflat/trace/illumflat
• INCAN=off → not typed as a flat/trace

Docs

• New "Setups and Frame Typing" section in doc/spectrographs/mmt_binospec.rst
• Release-note entries in doc/releases/2.1.0dev.rst

🤖 Generated with Claude Code

[tepickering]

This PR closes #2137.

[kbwestfall]

Looks good to me. We should run tests...

[tepickering]

Looks good to me. We should run tests...

i'll run a local dev-suite on mmt_binospec overnight. this PR does include unit tests of the changed functionality. to really exercise the changes requires a pypeit_setup vet test with some appropriate test data...

[bpholden]

Charlie has contributed four test frames: https://www.dropbox.com/scl/fo/0s19fvi3plzwngtou6k11/AGZ-vft3kA01hjLFmQU3zEE?rlkey=4mbvd90iqrqrbukkp4gbvvs0f&st=lpba3j97&dl=0 see: #2137

…

On Thu, Jun 4, 2026 at 9:40 PM T. E. Pickering ***@***.***> wrote: *tepickering* left a comment (pypeit/PypeIt#2142) <#2142 (comment)> Looks good to me. We should run tests... i'll run a local dev-suite on mmt_binospec overnight. this PR does include unit tests of the changed functionality. to really exercise the changes requires a pypeit_setup vet test with some appropriate test data... — Reply to this email directly, view it on GitHub <#2142?email_source=notifications&email_token=AAJJHGWH25RQHBGLBFQLT6L46JFLLA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTINRSHAZDKMBRGQYKM4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#issuecomment-4628250140>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJJHGXQVH4YIRKHLS4FFPT46JFLLAVCNFSM6AAAAACZWKTK2SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DMMRYGI2TAMJUGA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[tepickering]

grabbed the provided files and ran pypeit_setup on them. the fixes now create two configurations, one for each slit. the blank image with all lamps off is now commented out.

mmt_binospec_A.pypeit.txt
mmt_binospec_B.pypeit.txt

@charliekilpatrick, check these out and make sure they match your expectations.

(i had to append a .txt extension to get them to upload to github)

[charliekilpatrick]

grabbed the provided files and ran pypeit_setup on them. the fixes now create two configurations, one for each slit. the blank image with all lamps off is now commented out.

mmt_binospec_A.pypeit.txt mmt_binospec_B.pypeit.txt

@charliekilpatrick, check these out and make sure they match your expectations.

(i had to append a .txt extension to get them to upload to github)

Thanks, this looks correct to me - there should be two setups with the "A" setup flat file ignored due to INCAN=off. I appreciate the PR for our future Binospec science!

[tepickering]

local dev suite for mmt_binospec passed.

% $PYPEIT_DEV/pypeit_test reduce -i mmt_binospec                                                      26-06-05 - 10:54:42
Running reduce tests.
Running tests on the following instruments:
    mmt_binospec

Reducing data from mmt_binospec for the following setups:
    Longslit_G600
    Multislit_G270
    Longslit_G1000

 0 passed/ 0 failed/ 0 skipped STARTED mmt_binospec/Multislit_G270 pypeit
 1 passed/ 0 failed/ 0 skipped PASSED  mmt_binospec/Multislit_G270 pypeit
 1 passed/ 0 failed/ 0 skipped STARTED mmt_binospec/Longslit_G600 pypeit
 2 passed/ 0 failed/ 0 skipped PASSED  mmt_binospec/Longslit_G600 pypeit
 2 passed/ 0 failed/ 0 skipped STARTED mmt_binospec/Longslit_G1000 pypeit
 3 passed/ 0 failed/ 0 skipped PASSED  mmt_binospec/Longslit_G1000 pypeit
--- PYPEIT DEVELOPMENT SUITE PASSED 3/3 TESTS  ---

Test Summary
--------------------------------------------------------
--- PYPEIT DEVELOPMENT SUITE PASSED 3/3 TESTS  ---
Total disk usage: 10.3644 GiB
Testing Started at 2026-06-05T10:55:05.546428
Testing Completed at 2026-06-05T12:45:06.231579
Total Time: 1:50:00.685151

[tbowers7]

This looks like a good defensive guard against almost-flatfields being included.

Approve.

I would suggest adding these data to the DevSuite and adding a pypeit_setup VET test, however, to guard against regression.

[tepickering]

This looks like a good defensive guard against almost-flatfields being included.

Approve.

I would suggest adding these data to the DevSuite and adding a pypeit_setup VET test, however, to guard against regression.

this brings up a point we should discuss at a tagup meeting: what is the threshold for adding new tests to the dev-suite versus doing unit tests? in this case, the code/logic changes are pretty minor and easily covered by the included unit tests. behavior against real instrument headers is covered by existing mmt_binospec dev-suite tests (which all pass locally, as mentioned previously).