Fix mage conditional

See merge request !404

Fix broken imports related to the __init__.py changes for /ws endpoints

See merge request !403

Use the pymysql library to connect to the db

See merge request !402

Fix broken server references from reworking __init__.py and imports

See merge request !401

Database Standardization

See merge request !385

Replaced databases library with sqlmodel library. Organized code to include a /factories and /models under /db to move the metadata code under /api. Removed the contents of __init__.py to be empty files to follow the working agreement and clean up imports.

Minor miscellaneous fixes

See merge request !400

Create and use default Absolute class/module values, rather than hard-coding `shift` or `valid` values

- LegacySNCL's get_channel() now requires `data_type` argument
- For some reason, specifying a location code was intended to behave differently than specifying a `data_type` in SNCL.py. I can't imagine why, but I basically reverted a change made several years ago in commit 4fab9678 when I recently forced 'variation' and 'R?' to be treated the same way by SNCL.py.

There is an ambiguity that exists between `data_type` and `location` in *SNCL.py. Ideally, we'd clean this up, but that involves many changes, and well beyond just the *SNCL.py files. These changes at least fix issues with geomagio scripts that popped up during testing.

Handles case of broken IAGA2002 file where the channel headers don't align perfectly with the expected columns. Our IAGA2002 files never did this, but some from others did.

Clean up to support and/or match geomagio scripts

See merge request !399

- `--ouput-scale-factor` for EdgeFactory and MiniSeedFactory
- `--ouput-sncl-mode` for EdgeFactory and MiniSeedFactory

- Remove unnecessary DED filtering; just copy to new sncls
- Trust *SNCL.py to map `HEZ` channel codes correctly (still uses `UVW` when applying calibrations to 10Hz volts and bins)
- Add `T5` to ObsRIO minute filtering

Update version metadata for new release

See merge request !398

Copy F (geomag) to F (legacy) fixed:

See merge request !397

The EdgeFactory (earthworm client) does not have, and will probably never have, the ability to read floating point miniseed blocks. When we reverted the processing recently to stop making an integer copy of the 1-second data from ObsRio (which seriously confused Edge for reasons we still dont understand), this broke geomag-filter. The solution was to copy F data separately, using MiniSeedFactory as the input factory.

Do NOT "duplicate" raw 1-s data in EdgeCWB

See merge request !396

I mistakenly thought that only the float32 1-second data from ObsRIO systems was pushed to our ringserver, but apparently any/all LFF, LK1-4 channels are. This resulted in duplicate miniseed data going to IRIS, with one copy in the original float32 format, and another as scaled integers. This commit fixes this until/unless we can better reconfigure EdgeCWB to our needs.

Process default sncl_mode better:

See merge request !395

This was causing issues with the Controller, which always passes `sncl_mode`, even if it is None.

Add sncl_mode and scale_factor options to controller:

See merge request !394

I knew this in the previous commit, but I thought their parent class, TimeseriesFactory, could handle arbitrary keywords (i.e., it had a kwargs argument to vacuum up unrecognized keywords). I was mistaken.