Add method to find objects of a given neurodata type

[h-mayorquin]

This PR tries to complete the work started on PR #1737. It implements a couple of the design points raised in the previous discussion:

• The method is defined on NWBContainer rather than on NWBFile directly, so it works on any sub-container (ProcessingModule, LFP, etc.) and is correctly scoped to that container's subtree.
• String-to-class resolution goes through pynwb.get_type_map(copy=False), the live process-wide TypeMap (available since Remove unnecessary calls to deepcopy(get_type_map) #2121), rather than self.io.manager.type_map; the previous approach required the container to have been read or written through an IO, captured a snapshot of the TypeMap at IO-construction time, and so failed on in-memory containers and missed extensions loaded later in the same process.

Fix #560

Checklist

• Did you update CHANGELOG.md with your changes?
• Have you checked our Contributing document?
• Have you ensured the PR clearly describes the problem and the solution?
• Is your contribution compliant with our coding style? This can be checked running ruff check . && codespell from the source directory.
• Have you checked to ensure that there aren't other open Pull Requests for the same change?
• Have you included the relevant issue number using "Fix #XXX" notation where XXX is the issue number? By including "Fix #XXX" you allow GitHub to close issue #XXX when the PR is merged.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.19%. Comparing base (9231882) to head (ee0b91c).

Additional details and impacted files

@@           Coverage Diff           @@
##              dev    #2189   +/-   ##
=======================================
  Coverage   95.18%   95.19%           
=======================================
  Files          30       30           
  Lines        2991     2997    +6     
  Branches      444      445    +1     
=======================================
+ Hits         2847     2853    +6     
  Misses         86       86           
  Partials       58       58           

Flag	Coverage Δ
integration	72.93% <28.57%> (-0.12%)	⬇️
unit	85.61% <100.00%> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[oruebel]

@h-mayorquin this looks like something we probably would want to add on Container in HDMF instead since this is generic functionality that is independent of NWB specifics

[h-mayorquin]

@oruebel

The original PR supports matching both by class and by string (e.g. collect_descendants_of_type("ElectricalSeries")). For matching the strings we need the utility pynwb.get_type_map(). Naively moving this to hdmf would create a circular dependency, as hdmf.common.get_type_map() does not have all the mappings that pynwb.get_type_map() has (e.g. "ElectricalSeries").

We could support the basic matching with class only in hdmf and then a method here can use that functionality and add the string matching, concretely:

# hdmf (no nwb type map knowledge)
def collect_descendants_of_type(self, type_):
    return [obj for obj in self.all_children()
            if obj is not self and isinstance(obj, type_)]

# pynwb (add string resolution support)
def collect_descendants_of_type(self, neurodata_type):
    if isinstance(neurodata_type, str):
        neurodata_type = self._get_type_map().get_dt_container_cls(neurodata_type)
    return super().collect_descendants_of_type(neurodata_type)

One thing that I don't like is that the hdmf loses support for the pattern that @rly described (filtering objects whose Python class was auto-generated from a cached extension namespace at read time), and we have two slightly different functions in pynwb and hdmf which might be surprising and confusing. This compels me to ask: besides the logical place for a thing to be, what would be the concrete gain of having a reduced version on hdmf?