Configurable debouncedelay, increasing for polygon and arbitrum

[pragmaxim]

On fast chains like POL or ARB :

• there are frequent short reorgs that can make the block hash non-canonical before the trace completes (the -32603 error)
• traces can take longer than rpc_timeout allows, especially for large blocks

1. Increasing the delay before newHeads event coming and getting the actual traces and logs can make syncing smoother, waiting 500 extra Milis for each block is probably better than hanging or retries because of frequent Reorgs.
2. Passing explicit and increased trace_timeout to heavy debug_traceBlockByHash calls