LSI correlator: reconnect [Timeboxed: 5]

At the moment to reliably reconnect the LSI correlator, after fiddling with parameters on the local laptop, one needs to:
- Stop the IOC
- Power-cycle the box
- Wait ~1 minute
- Start the IOC

Otherwise, the LSI correlator will mysteriously fail to connect to the underlying websocket.

This needs to be made nicer - this is a horrible UX for use-cases where an experiment is swapping between control via IOC and setup on laptop.
- Are we misusing the vendor API / not closing old connections etc?
- If it can't connect immediately could it try waiting 60s then try again?

For the avoidance of doubt, the connection gets _physically_ swapped between an equipment-specific laptop with manufacturer software on it, and a network port on the main ISIS network which the IOC talks to. It is never connected to both.

This ticket is almost certain to need access to the hardware to test with. I believe the correlator can be used without the laser itself (it'll generate nonsense data) but check with scientist. Worst case, the laser can be set up offline.

This requires the real device so whoever does this ticket will need to organise some testing time.

discussed in planning 12/06/25 at 00:18:00

Start with the LSI correlator IOC's connection handling and reproduce the laptop-to-network-port swap described in the issue, observing the vendor websocket lifecycle. Confirm the failure and test whether reconnecting works without stopping the IOC or power-cycling the box; coordinate access to the correlator and scientist before testing.

Scritto dal modello di indicizzazione a partire dal testo della issue.