Implement ABBA-based analysis for chop-and-nod observations

[hapipapijuris]

Add ABBA method code #220
Should I add a scan time lapse plot like the figure outputted using pswsc function?

[hapipapijuris]

• Clean up ABBA method code
• delete unnecessary comment
• incorporate "abba" into all and Fire() and func auto

[astropenguin]

Thank you for implementing this feature. Your code assumes that the string abba exists somewhere in an observation name but this is not correct. I think you should update pswsc function instead of making a new one.

Should I add a scan time lapse plot like the figure outputted using pswsc function?

For the reason above, we don't need to make a new plotting code but simply replace the current spectrum made by the ABAB method with that by the ABBA method.

[astropenguin]

Minor comments:

1. Why is .data used for abba_cycle but not for abba_phase?

   abba_cycle = (scan_onoff.data * 2 + is_second_half - 1) // 4
   abba_phase = (scan_onoff * 2 + is_second_half - 1) % 4

2. The variable name da_abba_select does not look reasonable because it does not select the data anymore.
3. Please keep the plot format align with other functions. For example, the figure size of 6 x 4 inches is not common.

   fig, ax = plt.subplots(figsize=(6, 4))

4. I think required_elements = {0, 1, 2, 3} should be defined as a constant outside the function with the name of something like ABBA_PHASES.

[hapipapijuris]

• incorporate ABBA-based analysis code into pswsc func
• change name da_abba_select to spec
• make ABBA_PHASE in constant
• change scan_onoff to scan_onoff.data

[hapipapijuris]

I will add changing nod pair code in this issue. So please do not close.

[astropenguin]

Thank you for the update. Here are minor comments:

1. Please sort constants alphabetically and use a plural form for the ABBA phases.
2. Please add subtract_per_abba_cycle above subtract_per_scan (i.e. sort alphabetically).
3. Description of subtract_per_abba_cycle looks not correct.
4. Please rename subtract_per_scan because now it is not used per scan but per ABBA phase.
5. ABBA_PHASE == set(...) (L1177) certainly works but it is not a good coding practice: A comparison (a constant to be compared) should usually put on the right-hand side. In addition, .values does not seem necessary in this case.

[astropenguin]

Related to 5., it would be great to check out a coding style called "early returns" or "guard clauses". For example, subtract_per_abba_cycle following this style should become something like:

def subtract_per_abba_cycle(dems: xr.DataArray, /) -> xr.DataArray:
    if not set(np.unique(dems.abba_phase)) == ABBA_PHASES:
        return xr.DataArray(np.nan)

    return (
        dems
        .groupby("abba_phase")
        .map(subtract_per_abba_phase)
        .mean("abba_phase")
    )

[hapipapijuris]

I changed following Taniguchi-san's comment and reformatted the code.

[hapipapijuris]

change explanation of subtract_per_abba_cycle and data type scan_onoff.data -> scan_onoff

[ShinjiFujita]

I did not find any problems.

[astropenguin]

@ShinjiFujita san, thank you for your checking. I will merge the PR and make a new public release soon. By the way, my apologies for keeping your PR #219 still open... I will check it soon as well.