Consistent Metadata with updated versions of MTH5 and MT-metadata

[kujaku11]

Going through and trying to make sure metadata and methods are consistent with updated versions of MTH5 and mt-metadata before I merge them with master. The update was creating a single metadata object under the hood for mth5.timeseries.ChannelTS and mth5.timeseries.RunTS to allow for more self-consistent metadata. To do this an object was created for the list of metadata objects at the level below called mt_metadata.utils.list_dict.ListDict, which can be indexed as a list or keyed like a dictionary.

For example:

from mt_metadata.timeseries import Station, Run

s = Station(id="mt001")
s.add_run(Run(id="a"))
s.add_run(Run(id='"b"))

run_indexed = s.runs[0]
run_keyed = s.runs["a"]

# can be mixed slice as well
run_mixed = s.runs[0:"b"]
run_mixed = s.runs["a":]

Tasks Before Merging

• Go through and update metadata and methods to be consistent
• Make sure all tests run for all packages
  • MT-metadata
    • Merge or edit branch in MT-metadata PR#118
  • MTH5
  • Aurora

[codecov]

Codecov Report

Merging #239 (1037175) into main (0f7c8f8) will decrease coverage by 0.08%.
The diff coverage is 69.84%.

@@            Coverage Diff             @@
##             main     #239      +/-   ##
==========================================
- Coverage   77.62%   77.53%   -0.09%     
==========================================
  Files         101       99       -2     
  Lines        5470     5493      +23     
==========================================
+ Hits         4246     4259      +13     
- Misses       1224     1234      +10     

Impacted Files	Coverage Δ
aurora/pipelines/process_mth5.py	98.11% <ø> (ø)
aurora/pipelines/run_summary.py	88.88% <ø> (ø)
aurora/pipelines/time_series_helpers.py	75.57% <ø> (+2.38%)	⬆️
aurora/sandbox/butterworth_filters.py	0.00% <0.00%> (ø)
aurora/time_series/windowed_time_series.py	50.74% <28.57%> (-3.36%)	⬇️
tests/parkfield/test_process_parkfield_run_rr.py	78.26% <41.66%> (-16.19%)	⬇️
tests/parkfield/test_process_parkfield_run.py	82.25% <47.36%> (-13.58%)	⬇️
tests/parkfield/test_calibrate_parkfield.py	80.00% <61.11%> (-13.94%)	⬇️
aurora/sandbox/io_helpers/make_mth5_helpers.py	91.66% <75.00%> (-6.01%)	⬇️
aurora/test_utils/parkfield/make_parkfield_mth5.py	81.08% <81.08%> (ø)
... and 11 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[kkappler]

The aurora tests are failing for several reasons, which all appear to be because NCEDC is down, but none actually are.

• Python 3.6 failure is not fully debugged, but is likely due to a combination of issues that will be described in python 3.7, 3.8, 3.9
  • I suggest we deprecate support for 3.6 in aurora. The main reason for this is that a full python3 version of obspy are now available, (1.3, 1.3.1 and 1.4) but the 1.3 versions are python>=3.7 and the 1.4 version is python>=3.8.
• The python 3.7 failure was triggered by two things:
  • First, the builds on github-actions tests would have been using obspy v1.3 (or 1.3.1). This maps base_url "NCDEC" to an http address, not an https. Details in Aurora Pipeline Fails when NCEDC is down #159 comments Dec 29, 2022.
  • Second, there is an issue with an Auxiliary channel needing a component. Details in Aurora Pipeline Fails when NCEDC is down #159 comments Dec 29, 2022. This error is the same one that is breaking python v3.8 & v3.9 tests.

So, the strategy for getting this PR across the line should probably be to review the Auxiliary channel issue. I created an issue#117 in mt_metadata to take a look at this.

A few other observations of things that can be improved when this PR is merged:

• Add logic for NCEDC client so that if obspy Client does not initialize with base_url keyword "NCEDC", try agian with explicit address "https://service.ncedc.org/"
• get_response_inventory_from_server could actually be moved out of xml_sandbox and into fdsn_dataset_config
• Client can then be made an attribute or method of the FDSNDatasetConfig class, and then we don't need to reinitialize it when getting data
• in mth5 the read_back_data method should close the mth5 object before returning