Major refactor/improvement in command interrupt handling

[taldcroft]

Description

This implements an idea that has been around for a long time to handle command interrupt processing in a more direct -way that more closely matches the OBC behavior. In particular:

• Notice END SCS commands (CODISASX) and cut all running commands in that SCS slot after the command time.
• Implement RLTT as END SCS for slots 128 to 133.
• Put appropriate END SCS commands in the Command Event definitions.

A key implementation detail is the definition of "running commands". The basic outline of how this works is as follows.

• Make a list of CommandTable objects corresponding to each approved command load
• Append to that list a CommandTable of commands for each Command Event.
• Sort the list of CommandTables (called cmds_list) by the date of the first command in each table.
• Iterate over cmds_list (for cmds in cmds_list:)
  • Find any RLTT or END SCS commands in cmds and collect them into a dict that captures the SCS numbers and end dates.
  • Iterate over all previous cmds in the cmds_list and cut commands accordingly.
• Stack the cmds_list tables and then sort into the standard backstop order.

The key point to this change is that it simplifies the code logic and removes ad-hoc processing rules.

To do:

• Confirm that loads.csv is no longer used as an input to processing.
• Consider further cleanup in code and maybe remove the loads.csv and loads.dat output.
• Check command history near 2022:223 BSH. Dither stays enabled?

Interface impacts

New END SCS and LOAD_EVENT commands are stored in the archive. No changes to actual spacecraft commands.

Testing

Unit tests

• Mac with current "flight" archive ($SKA/data/kadi/cmds2.*)
• Mac with new archive created using this code.

Independent check of unit tests by Jean

• Linux

Functional tests

• Regenerate full-mission commands archive and confirm only expected diffs. See validate/validate-cmd-interrupt-refactor.ipynb