Fix data gap

[pavlis]

This started out as a simple bug fix for a mistake in the pybind11 code for this auxiliary class. I realized, however, that in working with a prototype for surgical mutes that this was a very useful container that might work well as a component of a generic surgical mute algorithm. Working on the prototype with a data set from Homestake I realized it lacked a couple important capabilities:

1. The previous version required gaps intervals to not overlap. An attempt to add a gap definition that overlapped an existing one would cause nothing to happen. That happened because the original concept was recorded data gaps, which are effectively always immutable. For a more generic use that needed to be relaxed. I did that here by adding some minor complexity to the add_gaps method. Tests show it that functionality now works.
2. I saw a need for the C++ operator += (need for the using += in python of course) that would cleanly merge two different DataGap containers. Once functionality 1 was added this was trivial. Well, not really as there was an odd property of the std::set container that complicated the code requiring a complete copy of the left hand side before merging the right hand side. What I did works, but I'm not quite sure why it was necessary. I considered it harmless, however, as this container should rarely be huge enough that such a copy operation is a serous performance problem.

[codecov]

Codecov Report

Attention: 6 lines in your changes are missing coverage. Please review.

Comparison is base (23b0dc5) 55.01% compared to head (7d7cc0a) 55.08%.

Files	Patch %	Lines
cxx/src/lib/seismic/DataGap.cc	90.00%	4 Missing ⚠️
cxx/include/mspass/seismic/TimeSeriesWGaps.h	0.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #508      +/-   ##
==========================================
+ Coverage   55.01%   55.08%   +0.07%     
==========================================
  Files         144      144              
  Lines       22382    22429      +47     
==========================================
+ Hits        12313    12356      +43     
- Misses      10069    10073       +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[pavlis]

Looks like these changes all passed the tests clean as I expected. Wanted to comment that the lines Codecov flags are harmless. They are present because I intentionally omitted the list list of TimeWindow constructor. It was a relic of the older seispp api and seemed less useful for the python api. If you, @wangyinz, think it would be useful thing to add it wouldn't be hard. More important is the TODO you will see the pybind11 code - this should really have a pickle api. Won't really need that, however, unless I decide to move forward with this prototype surgical mute using this as the container. Until then I'd suggest this thing is kind of a niche component.