Fix panic in ring source

[seydx]

In my previous PR, I forgot to correctly execute MarshalJSON on a snapshot producer. As a result, for example, the /probe call didn't work on it.

Also, during my testing, I noticed that in rare cases the following error occurs:

DBG [streams] start producer url=ring:?device_id=XXX&refresh_token=XXX&snapshot
DBG [streams] stop producer url=ring:?device_id=XXX&refresh_token=XXX&snapshot

[signal SIGSEGV: segmentation violation code=0x2 addr=0x28 pc=0x102d10330]

goroutine 891 [running]:
github.com/AlexxIT/go2rtc/pkg/ring.(*Client).Start(0x5?)
	github.com/AlexxIT/go2rtc/pkg/ring/client.go:504 +0x20
github.com/AlexxIT/go2rtc/internal/streams.(*Producer).worker(0x140004a2cf0, {0x102fd9128?, 0x1400047a540?}, 0xd)
	github.com/AlexxIT/go2rtc/internal/streams/producer.go:161 +0x34
created by github.com/AlexxIT/go2rtc/internal/streams.(*Producer).start in goroutine 879
	github.com/AlexxIT/go2rtc/internal/streams/producer.go:157 +0x1b0

This PR should fix this by no longer setting the producer to nil during a stop (as is the case with other producers)

[AlexxIT]

Thanks!

[AlexxIT]

https://github.com/AlexxIT/go2rtc/releases/tag/v1.9.9