Two new functions to handle limit strings, minor viz fixes for xija_gui_fit

[jzuhone]

Description

Based on the new limit scheme implemented in sot/chandra_models#75, this PR implements two new functions in xija, get_limit_spec and get_limit_color.

get_limit_spec takes a limit of the form <system>.<type>.<direction>.<qualifier> and returns a dict with these names as the keys:

from xija.limits import get_limit_spec
limit_spec = get_limit_spec("planning.data_quality.high.acisi")
print(limit_spec)

which gives

{
    'system': 'planning', 
    'type': 'data_quality', 
    'direction': 'high', 
    'qualifier': 'acisi'
}

This may have other uses eventually, but it was useful for get_limit_color, which uses it to assign colors for the various limits.

get_limit_color is used by xija_gui_fit to assign limit line colors, but I'll be using and get_limit_spec and get_limit_color in acis_thermal_check and ACISpy.

Functional testing consisted of running xija_gui_fit for every model in chandra_models with limits and making sure that the limit lines appeared correctly in the data__time, resid__data, and histogram plots.

This PR also fixes a number of minor visualization issues with xija_gui_fit.

Testing

• Passes unit tests on MacOS, linux, Windows (at least one required)
• Functional testing

Fixes #

[jzuhone]

@taldcroft Wanted to see if you had any comments on this.

[taldcroft]

I think this should pay attention to the limit units (indeed many of them are not degC).

[jzuhone]

@taldcroft what do you want it to do exactly? This loop is to draw the lines and since the "unit" item is in the same dict as the limit definitions, it needs to skip over it which is the reason for lines 170-171.

[taldcroft]

E.g the limits for the PFTANK2T model are in degF, but the models are always computed and calculated in degC. Maybe I'm missing the code, but somewhere along the way the limit values need to be converted degC.

[jzuhone]

This is ready for review again.

[taldcroft]

E.g the limits for the PFTANK2T model are in degF, but the models are always computed and calculated in degC. Maybe I'm missing the code, but somewhere along the way the limit values need to be converted degC.

[taldcroft]

Looks good now!