feat: add event dataset browser with physics-aware filtering (#861)

[rx18-eng]

Summary

• added a new Event Browser panel that pre-scans all loaded events and shows a summary table with object counts per collection type, MET, and event/run numbers
• the table is sortable and filterable (column filters with >=, <=, =, MET minimum, and search by event number)
• keyboard nav works with arrow keys in the table + Shift+Left/Right for prev/next event globally

columns are ordered with reconstructed physics objects first (Jets, Muons, Tracks, etc) and detector-level stuff (CaloCells, Hits) pushed to the right. zero values are dimmed so you can quickly scan which events have what. there's a separate "Reco" column that only counts reconstructed objects for easier comparison between events.

follows the same CDK overlay pattern as Collections Info, config-driven via showEventBrowser in UIMenuConfig.

search by event number is there per Edward's feedback. the UI is independent of the loading strategy so lazy-loading can be added later without changing the browser itself (per Sebastien's note on memory).

Test plan

• load a multi-event file (atlas.json from assets/files/event_data/ works)
• open Event Browser from toolbar
• check that reconstructed columns show before detector-level ones
• try sorting, adding filters, searching by event number
• click load button and check the spinner + loaded event highlight
• Shift+Right/Left to navigate events globally

Recording.2026-03-31.012641.mp4

[rx18-eng]

@EdwardMoyse @sponce , please let me know your thoughts on this !

[sponce]

I like this a lot. And I've just tested it with LHCb data, it works very nicely. Just a tiny question : what is the "MET" column and why is it always empty ?

[rx18-eng]

@sponce Thanks for your review . MET is Missing Transverse Energy, computed from the MissingEnergy collection if the event data has one (e.g. ATLAS JiveXML files). LHCb data doesn't include MET collections since it's a forward spectrometer, so the column stays empty.

IF you want I can hide the MET column entirely when no events have MET data so it doesn't show up as noise for experiments that don't use it.

[sponce]

Thanks for the explanation about MET. Coming from LHCb, I must say I did not know the concept indeed ! Indeed, the columns not appearing at all in data should probably not be displayed, although it's not a big deal. The feature is already very useful like this in my opinion