Daemonized version for watched folders etc.

[ml-evs]

I'm wondering if it is worth adding a daemonized version of the Python API functionality for watching folders and pushing to samples (for example). It could either live in this repo or somewhere separate, but perhaps as it will use this repo heavily, it could be prototyped here as an example. I'm not sure who has a use case for this yet (@PeterKraus / @jdbocarsly?) but I imagine it will be common (needing to push from instruments rather than pull).

[ml-evs]

Expanding on this a bit with a concrete use case.

"Multi-tenant" datalab for a consortium <==> instrument in one group's lab

They don't want to expose the instrument directly to datalab as a FS mount, but the instrument machine can see the internet (or at least backs up to a place that can) Instead, need a daemon on the instrument machine that:

1. watches a folder
2. sends fs metadata to datalab (ideally as deltas), on a schedule or on change (i.e., on every file change but at most every n mins).
3. instrument is configured with an owning group who can see it as a possible remote FS (should also be tied to a datalab equipment entry) with its own API key
4. user can request to sync a file from the remote fs as currently happens, but instead of actually doing the sync, datalab serves an endpoint for the instrument that lists the files the user wants
5. on a different schedule (e.g. every minute) the daemon polls that endpoint and pushes file contents directly to datalab, attaching it to the sample that the user requested