Allow for empty "properties" subdict

[teutoburg]

Fixes a long-ignored bug where some keys could "get lost" in the __currsys__ if an "alias" but no "properties" is present in a yaml dict. This occurs e.g. in MICADO's default.yaml in the SCAO and MCAO modes, that don't (and probably don't need to) define any "properties". But then NestedMapping.update() defaults to simply updating key-value-wise.

Devil's Advocaat

The alternative would be to introduce an empty "properties" in each of the yamls. But then people will forget that somewhere and we're back at the bug, so that's fragile. Also would probably require an update to multiple IRDB packages, which is annoying.

Rant

I'm generally not a fan of this "abusive" overloading of the .update() method. I think that method should only do what you'd expect from a standard dict (in fact, by inheriting from collections.abc.MutableMapping, we wouldn't even need to define such a "standard" .update() method here...). The "special sauce" that's added here is also rather ScopeSim-specific and thus shouldn't really be in astar-utils to begin with. Perhaps an extra method (.update_alias()??) would be fine though. Same thing actually goes for UserCommands.update() over in ScopeSim, although I started to mitigate that in the (as of writing this un-PR'd) fh/chainmap branch...

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (445264f) 99.04% compared to head (ce08c1e) 99.05%.
Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #32   +/-   ##
=======================================
  Coverage   99.04%   99.05%           
=======================================
  Files           5        5           
  Lines         315      316    +1     
=======================================
+ Hits          312      313    +1     
  Misses          3        3           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[hugobuddel]

Good! Sorry for taking so long to review.

It seems reasonable to have this alias and properties stuff somewhere if that is useful to ScopeSim, since that is the main reason we have these classes right?