Update sot/xija to 4.19.2

[javierggt]

Includes the following merges:

• PR #132: Update location of chandra_models readme
• PR #131: Add xija.component.heat to installed packages in setup.py
• PR #129: Refactor heat.py into a subpackage
• PR #130: Fix problem of dropping bad_times in spec file and y-scale divergence
• PR #123: Improve performance of bad_times and mask_times
• PR #121: Updated 'Hack' to only overwrite final prediction mvals instead of all mvals + PEP8
• PR #117: Adjust earthshine effect based on Earth phase
• PR #119: Fix epoch interpolation bug, wrong function name for interpolation
• PR #116: Updates for xija_gui_fit
• PR #115: remove use of deprecated np.float and np.int
• PR #113: Fixing bugs in limits implementation and adding them to the model info window in xija_gui_fit
• PR #112: Two new functions to handle limit strings, minor viz fixes for xija_gui_fit
• PR #111: Use the right repo path to get available model names
• PR #110: Set the default repo_path inside the get_model_spec function
• PR #109: This fixes a bug that was missed in the recent gui_fit PR
• PR #108: Fix random crashes using xija_gui_fit on Linux, change limit names for annotation, use CxoTime
• PR #107: Allow fewer dP pitches + add sensible keyword arg defaults
• PR #106: Add a Delay model component
• PR #104: Use github's 'latest' release link for latest version
• PR #103: Add MSIDStatePower component
• PR #102: Fix get_model_spec to work on Windows
• PR #100: Add timeout error catch that occurs on linux
• PR #96: Add module to allow getting model spec files from Ska data
• PR #99: Allow model_spec to be provided as Path
• PR #97: Migrate all doc strings to use numpydoc
• PR #95: Unified documentation style
• PR #93: Shiny workflow update
• PR #92: Shiny workflow update
• PR #91: Shiny