Method to load Bench should be calibration?

[monodera]

Yabe-san's message on a Slack DM

I found that the way to load the bench information is different between PPP and SFA.

In PPP,

bench = Bench(layout="full")

In SFA (actually in nfutils.py of ets_pointing),

bench = Bench(
        layout="calibration",
        calibrationProduct=calibrationProduct,
        blackDotsCalibrationProduct=blackDotsCalibrationProduct,
        blackDotsMargin=black_dot_radius_margin,
    )

In my current branch of pfs_obsproc_planning_tools, I use the latter using nfuitls of ets_pointing package. The difference may be small but I think we should use the consistent way of bench even in online PPP. The implementation can be found here: https://github.com/Subaru-PFS/pfs_obsproc_planning_tools/blob/u/kiyoyabe/e2e_test_2023oct/src/pfs_obsproc_planning/PPP.py
(You may also need to look at other codes...)
Also, if you have time (or you have also done some updates for full PPP), could you check the current PPP.py (above one) please?

By the way, fixing this discrepancy improves the discrepancy between PPP output and SFA output. There is still a slight difference even if calibration sources are NOT included, which I still cannot understand.