If a user has specific Bluesky runs that can inform their optimization objectives, we should allow them to supply us a UID and we can unpack it, compute the objective values, and attach it as a trial in Ax.
This way, Ax's modeling, suggestions, and analyses can utilize this data even though it was collected prior to experimenting with Blop or Ax.
See https://ax.dev/docs/recipes/existing-data