[aggregator] Add M3Msg client and server for M3Aggregator

[robskillington]

What this PR does / why we need it:

This adds the ability for users to use M3Msg to write and consume.

Special notes for your reviewer:

Does this PR introduce a user-facing and/or backwards incompatible change?:

NONE

Does this PR require updating code package or user-facing documentation?:

NONE

[schallert]

Are all of the below error cases non-retryable? If we unconditionally ack we'll never retry from the coordinator client to the aggregator right?

[robskillington]

I believe we probably should avoid retries on this, if aggregator gets overloaded the amount of buffering/retries will probably quickly overwhelm the coordinators.

[codecov]

Codecov Report

Merging #2171 into master will not change coverage by %.
The diff coverage is n/a.

@@          Coverage Diff           @@
##           master   #2171   +/-   ##
======================================
  Coverage    71.9%   71.9%           
======================================
  Files        1024    1024           
  Lines       89726   89726           
======================================
  Hits        64581   64581           
  Misses      20827   20827           
  Partials     4318    4318           

Flag	Coverage Δ
#aggregator	77.6% <0.0%> (ø)
#cluster	84.8% <0.0%> (ø)
#collector	82.8% <0.0%> (ø)
#dbnode	79.2% <0.0%> (ø)
#m3em	74.4% <0.0%> (ø)
#m3ninx	74.4% <0.0%> (ø)
#m3nsch	51.1% <0.0%> (ø)
#metrics	17.5% <0.0%> (ø)
#msg	74.9% <0.0%> (ø)
#query	68.2% <0.0%> (ø)
#x	83.1% <0.0%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 02d1877...48beec5. Read the comment docs.

[notbdu]

Left a few comments.

Changes overall LGTM but I don't have any context on things like why the legacy protocol has to watch for placements while the new m3msg protocol does not.