Improve doc and cleanup for selection_optional argument in estimators

[QRemy]

This PR fix the "all" option for in selection_optional argument of estimators and improve the docstring. Resolve issue #3295.

[codecov]

Codecov Report

Merging #3309 (04d7e5d) into master (b36cb89) will decrease coverage by 0.00%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #3309      +/-   ##
==========================================
- Coverage   93.83%   93.82%   -0.01%     
==========================================
  Files         144      144              
  Lines       17998    18002       +4     
==========================================
+ Hits        16888    16891       +3     
- Misses       1110     1111       +1     

Impacted Files	Coverage Δ
gammapy/estimators/excess_map.py	98.97% <ø> (ø)
gammapy/estimators/excess_profile.py	96.70% <ø> (ø)
gammapy/estimators/flux.py	92.95% <ø> (ø)
gammapy/estimators/lightcurve.py	89.80% <ø> (ø)
gammapy/estimators/parameter.py	98.57% <ø> (ø)
gammapy/estimators/ts_map.py	97.99% <ø> (ø)
gammapy/estimators/core.py	94.79% <100.00%> (ø)
gammapy/estimators/flux_point.py	93.67% <100.00%> (+0.72%)	⬆️
gammapy/modeling/iminuit.py	92.39% <0.00%> (-3.27%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b36cb89...04d7e5d. Read the comment docs.

[AtreyeeS]

Thanks a lot @QRemy ! I think this PR still needs to homogenise between scan and norm-scan keywords. Also, as I wrote in #3295, it would be helpful to for a user to have a proper message in .plot_ts_profile if scan is not execute - maybe searching for the column and throwing an error is useful, instead of incomprehensible message we have now

[QRemy]

@AtreyeeS, the message I see for FluxPoints.plot_ts_profiles() is"Missing columns for sed type '{}':" " {}".format(sed_type, missing) which is quite explicit. Do you have an example to reproduce the odd error message you have ?

[QRemy]

I added a check for optional arguments missing in ._validate_table(table, sed_type, use_optional=False).

[AtreyeeS]

Thanks again @QRemy ! No further comments from my side

[adonath]

Thanks @QRemy and @AtreyeeS for the review. I have no further comments either...