Pipelining Followups

## Problem

Some issues remaining at the end of #128. However, the long validation process made it safer to push fixes as followups

## Remove the `*pImpl` parameter from `Retries()` in `conduit/pipeline/common.go`
Per @winder
> * breaking out the context.Context out into a parameter, I was specifically looking for it to make sure the wait is properly terminated.
> * also breaking out RetryCount and logger. It's already a pure function in how it's used, but the signature doesn't guarantee that it's pure.

### Alternatively, just make `Retries` private

## Trim the special end of round message
In the final EC2 test, the size of the log at catchup grew from 14.1 GB to 16.2 GB. This is attributable to the change of the end of round log message. Currently it looks like

```txt
UPDATED Pipeline NextRound=31446285. FINISHED Pipeline round r=31446284 (6 txn) exported in 3.721356ms
```
In particular the `UPDATED...` prefix was added.

While this message requires a special format to comply with our testing (internal tools `logstats.go` and the E2E test), we can certainly trim it down and the UPDATED `NextRound` info is not crucial (it can be carved out as a `DEBUG` level message if not already present).

## Urgency

Medium