Xiaoya fix xps

[xiaoyachong]

This PR is used to fix issues encountered during XPS beamtime on Oct 23:

1. New tiled data organization for live results of arroyo_xps

{tiled_base_uri}/{tiled_prefix}/
│
└── {root_segments}/
    └── {USER}/
        └── {YYYY}/                    
            └── {MM}/                 
                └── {DD}/             
                    └── {experiment_name}/
                        └── {UUID}/
                            ├── feature_vectors (table for saving feature vector results)
                            └── xps_averaged_heatmaps (3D array for saving heatmaps)

Structure Breakdown:

• Base URI: {tiled_base_uri}/api/v1/array/full/{tiled_prefix}/
  Example tiled_base_uri: https://tiled.computing.als.lbl.gov
  Example tiled_prefix: /beamlines/bl931/processed/

• Image Data (3D Arrays):
  Query parameters: ?slice={index}:{index+1},0:{height},0:{width}
  Example tiled_url for the 40th image in a scan: http://tiled:8000/api/v1/array/full/beamlines/bl931/processed/lse_live_results/xiaoyachong/2025/10/28/TRS_test1/f40f7aec-1f26-4273-ac57-2def3906ac12/xps_averaged_heatmaps?slice=39:40,0:983,0:150

Note: We use write_array and patch to save image data to Tiled, as shown in this reference. The latest version of Tiled is required to support the patch method.

• Feature Vectors (Tables)

This change addresses #69

2. Add missing packages for arroyo_xps in .toml
3. Add time-based coloring and date picker in experiment replay

[dylanmcreynolds]

How do tiled_prefix and root_segments differ?

[xiaoyachong]

Example tiled_prefix: /beamlines/bl931/processed/
Example root_segments: lse_live_results

The tiled file structure is like:

{tiled_base_uri}/{tiled_prefix}/
│
└── {root_segments}/
    └── {USER}/
        └── {daily_run_YYYY-MM-DD}/
            └── {experiment_name}/
                └── {UUID}/
                    ├── feature_vectors (table for saving feature vector results)
                    └── xps_averaged_heatmaps (3D array for saving heatmaps)

[dylanmcreynolds]

Thanks for the map, that's helps. I see now.

I wonder whether we should remove {USER}. Multiple users may want to browse the results, and the date seems to be the a more significant than who was taking it.

To keep one single collection from growing too large, what do you think of creating subfolders by year, month than day?

[xiaoyachong]

Thanks for the suggestion.

The {USER} component is currently used as a restriction in experiment_replay, ensuring that users can only replay their own experiments.
@taxe10 I’m not sure whether we should remove this — perhaps Tanny can share her thoughts.

Regarding the {daily_run_YYYY-MM-DD} container node, we could split it into three hierarchical layers:

{tiled_base_uri}/{tiled_prefix}/
│
└── {root_segments}/
    └── {USER}/
        └── {YYYY}/                    
            └── {MM}/                 
                └── {DD}/             
                    └── {experiment_name}/
                        └── {UUID}/
                            ├── feature_vectors (table for saving feature vector results)
                            └── xps_averaged_heatmaps (3D array for saving heatmaps)

[dylanmcreynolds]

Nice! A couple of changes about redis and async.

[xiaoyachong]

Nice! A couple of changes about redis and async.

Thank you for your comments! I addressed them in my latest commit.