[API] Improvements to error handling in SpikeGLX

[dbirman]

• It looks like if there’s an error in talking to SGLX – for example, if I try to match a 4-shank probe in Pinpoint with a single shank probe in SGLX – that I can’t connect to SGLX again without closing Pinpoint and restarting.
• If there’s a connection open to SGLX and I stop the SGLX run, I can’t reopen the connection to SpikeGLX from Pinpoint – sometimes there’s an error saying it can’t connect, sometimes it looks connected but isn’t sending any messages. Quitting Pinpoint and restarting fixes this issue.

From JC by email