Review use of `static` cache for VP sensor conditions.
The following discussion from !3134 (merged) should be addressed:
-
@graven started a discussion: Not related to this MR, but the use of a
static
here is clearly with the intent that cache is shared by all sensors -- which is OK for any given event time, as all sensors live in the 'same' event, but it is not guaranteed to be OK when there are potentially multiple runs 'in flight', i.e. when processing has hit a run boundary. Specifically, it depends on how processing 'crosses' the boundary, and whether there will ever be two sets of conditions concurrently required.So resolving this into an issue to raise awareness to this...