[K8sDnsNameResolver] Fix Netty panic logging on JVM shutdown

[migesok]

Removes this erroneous log message on JVM shutdown:

io.grpc.StatusRuntimeException: INTERNAL: Panic! This is a bug!

Why it was logged:

• it seems to happen when LookupSession.lookupAsync tries to establish a new DNS server connection after the JVM shutdown has already started (it could happen because JVM shutdown hooks are executed in unpredictable order)
• under the hood that function tries to register a JVM shutdown hook to clean up NIO resources registering a shutdown hook during the shutdown sequence results in java.lang.IllegalStateException: Shutdown in progress
• this exception is not reported as a failure case in LookupSession.lookupAsync result completion stage but is thrown directly on the method call
• the code inside the resolver doesn't handle such direct exception gracefully, which causes it to propagate to Netty internals, resulting in a scary "Panic! This is a bug!" log message written to JUL by Netty internals

The fix:

• catch and handle the "error on JVM shutdown" case
• reduce the log level for such errors below INFO
• additionally, the resolver shutdown logic was hardened for all the cases where currently running resolution is in a race with the shutdown

[fillson-shady]

Nice fix, thank you!