fix for subset handling -> display ID updates

[shouples]

Lingering issue after #135 was that jumping to different cell executions and same-dataset rendering would mix up display IDs.

The simplest solution is to just pop the hash out of the subset cache so the only time we add data is during a resampling event, which means the only time it's going to be checked is immediately after an update_display_id() call when format() is being called. Once that update process is done, we pop the hash out, and no follow-on confusion happens.

Extra: minor metadata updates to handle subsets pulling parent metadata to include resample filters.

[codecov-commenter]

Codecov Report

Merging #137 (85e9894) into main (c516580) will decrease coverage by 0.03%.
The diff coverage is 62.50%.

@@            Coverage Diff             @@
##             main     #137      +/-   ##
==========================================
- Coverage   80.13%   80.09%   -0.04%     
==========================================
  Files          48       48              
  Lines        2436     2442       +6     
==========================================
+ Hits         1952     1956       +4     
- Misses        484      486       +2     

Impacted Files	Coverage Δ
src/dx/plotting/dashboards.py	22.22% <0.00%> (ø)
src/dx/plotting/dex/basic.py	21.42% <0.00%> (ø)
src/dx/plotting/main.py	26.31% <0.00%> (ø)
src/dx/formatters/main.py	83.33% <57.14%> (-1.14%)	⬇️
src/dx/utils/formatting.py	87.11% <85.71%> (ø)
src/dx/filtering.py	100.00% <100.00%> (ø)
src/dx/types/dex_metadata.py	94.08% <100.00%> (ø)
src/dx/utils/tracking.py	95.60% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6e75b65...85e9894. Read the comment docs.