Python-level dead band

[danielballan]

Different experiments might call for different levels of precision in tracking changes in, say, the beam current. One user might want a detailed data dump, another might want a shorter table that only records larger shifts. (And all the detailed data is being captured in the archiver anyway.)

We can add a Python-level deadband in the monitor message or in the Device.subscribe method to enable this.

via a conversation with @cmazzoli

[tacaswell]

This is a cross opdyd/bluesky problem where we need to make sure that we can thread the configuration from the message down through ophyd.subscribe through to epics.

This probably also means we need to support at least dead-band filtering on Signal. Moving to ophyd as that is likely to be where most of the hard work is.

[prjemian]

Checking the ao record documentation, there is the 'MLST' field (Last Value Change Monitor Trigger Value) that is almost what issue describes. Changing this value changes when the PV monitors are triggered. Close, but prolly not close enough. We need a python-specific value?

[prjemian]

Oh, nevermind since MLST can't be modified from a CA client. At least we looked.

[danielballan]

There is renewed interest in this, prompted by @EliotGann.