Simplify CLI usage, debug slow dataflow, and fix UDS timeout problems

[nraultwang]

During our observing training on Oct 17, we noticed the following QoL and minor bugs with the daq_data gRPC code:

• Simplify cli usage (1/3): make --init server initialization only required if any filesystem polling data sources are activated. By default, the server only uses Unix-Domain Sockets (UDS) so this step is unnecessary.
• Simplify cli usage (2/3): don't require $DAQ_CFG $NET_CFG. Instead, automatically import them via os.environ.get('DAQ_CFG') in Python.
• Simplify cli usage (3/3): Update CLI documentation to reflect this simpler usage.
• Debug (1/2): When we run plot-view, the gattini grpc server takes 30s to a minute to start streaming data at a steady rate. Attempt to replicate this issue and debug the root cause.
• Debug (2/2): The gRPC server frequently closed and reopened the UDS socket file, possibly due to low data rate or other complication factors. Attempt to replicate this issue.

System environment for replication:

1. The Observing Software (pcap-based) was collecting data in parallel with Hashpipe.
2. Data config: 9.5 ph threshold, 2-pixel trigger -> slow event rate.
3. Observing with just Gattini.