Fix language in recommendations for ack-range limits

[janaiyengar]

In trying to do a simple fix for #3311, I found other irrelevant or poorly-written recommendations in how a receiver could limit ACK ranges it sends. This section needs to be rewritten.

Specifically, the sentence below is a poor recommendation and contradicts earlier recommendation on acking until the ACK frame is acknowledged.

   The receiver SHOULD
   exclude already acknowledged packets from future ACK frames whenever
   these packets would unnecessarily contribute to the ACK frame size.

The recommendation below is fine, but the rationale does not make sense:

   Standard QUIC algorithms ([QUIC-RECOVERY]) declare
   packets lost after sufficiently newer packets are acknowledged.
   Therefore, the receiver SHOULD repeatedly acknowledge newly received
   packets in preference to packets received in the past.

This should be a design change.

[ianswett]

I'm not in love with this (entire) section either, but I'd like to better understand this issue.

I agree that sentence is mis-written, though I think the intent was fine.

On the rationale, can you clarify why it does not make sense?