feat(timing): compatability with corrections generated in real-time

[ssiegelx]

• The TimingCorrection.from_acq_h5 method will now automatically
  recognize the chime_delay acquisition files that are generated
  by the real-time pipeline, and use a specialized method for
  generating a valid timing correction from these files.

• Add a predict_x_error method to the TimingCorrection class,
  which will predict the offset in the telescope x position of sources
  due to the measured timing variations.

[ssiegelx]

@danielemichilli This PR should enable you to load timing correction files generated by both the offline and real-time pipeline.

It also adds a predict_x_error method to the TimingCorrection class that you should use instead of the version in the baseband-analysis.localization module. It is called as:

dx = tcorr.predict_x_error(timestamp, input_map=input_map)

where input_map is the output of ch_util.tools.get_correlator_inputs. Note that if you do not provide the input_map argument, then the code will query the layout database for this information. This query is the slowest part of the routine, so if you are calling it many times you may want to pass it as an argument.

[danielemichilli]

@ssiegelx I was finally able to test this, sorry about the huge delay.

The new function formally works but the correction I get is different from the previous version.
For example, I am testing TOA = 1553730069.4198346138, are you able to doublecheck the correct value I should be getting for that timestamp?