fix: Replace mutex with channel for TelemetryCounter

[shazlehu]

Proposed Change

Users are seeing agents stuck at restart after reconfiguring, and my working hypothesis is that the agent is having trouble with a pipeline that includes 16 count_telemetry processors, each of which uses a mutex. The obvious problem is that the agent defers unlocking the mutex until the next consumer in the ConsumeLogs chain returns, but it seemed like a good idea to rewrite the processor using channels instead of mutexes, the preferred Go synchronization method

From a mutex profile of a stuck agent:

Checklist

• Changes are tested
• CI has passed