Enable PP upload.py for LSSTCam-ImSim

`upload.py LSSTCam-imSim` is now part of the test suite

Using 2.2i/defaults chain directly as I chained previously meant trouble about calibration collection 2.2i/calib versus LSSTCam-imSim/calib in the local repo, and it needed an extra calib collection export. Following the discussions at I re-chained the LSSTCam-imSim/defaults in the dev repo:

butler collection-chain s3://rubin-pp-dev-users/central_repo_2 LSSTCam-imSim/defaults LSSTCam-imSim/calib,skymaps,refcats,LSSTCam-imSim/templates,pretrained_models
This is more consistent with other instruments.

With the ticket branch from here and , `python upload.py LSSTCam-imSim 2` uploads 3 detector images from exposure 496960 and 496989, and all 3 images can finish ApPipe and generate alerts.

please review
https://github.com/lsst-dm/next_visit_fan_out/pull/21 and https://github.com/lsst-dm/prompt_processing/pull/255 . Thank you!

LSSTCam-imSim is weird because it's an LSST instrument but does not follow most LSST conventions in their headers. Forcing it to be really like LSST means changes in `obs_lsst` and may invalidate existing data, which we don't want. On the other hand PP will only process newly uploaded imsim data and we can overwrite headers for those.

Adding notes here as this could bite us later… As of 2025-02-11

The DC2_cells_v1 in /repo/dc2 is named lsst_cells_v1 in other repos; it has the same hash.

TheDC2_cells_v1 in all other repos, including s3://rubin-pp-dev-users/central_repo_2 , have a different hash. Those have a different tractOverlap in their config.

While imsim templates are generated in /repo/dc2 and now seems too late to correct it for . I think we can live with it somehow, just that it will likely generate confusion.

imsim DC2 /repo/dc2 has a different convention and have defaults collections under 2.2i/x , not LSSTCam-imSim/x . PP relies on those collection names, so I added chains in pp-dev repo: