Reduce Multi allocations

[jponge]

No description provided.

[jponge]

@cescoffier Note that this is still a draft 😆

[codecov]

Codecov Report

Merging #690 (fec1937) into main (cbfbfcf) will increase coverage by 0.20%.
The diff coverage is 76.22%.

❗ Current head fec1937 differs from pull request most recent head c98647e. Consider uploading reports for the commit c98647e to get more accurate results

@@             Coverage Diff              @@
##               main     #690      +/-   ##
============================================
+ Coverage     89.87%   90.07%   +0.20%     
- Complexity     2977     2993      +16     
============================================
  Files           374      374              
  Lines         11770    11791      +21     
  Branches       1476     1480       +4     
============================================
+ Hits          10578    10621      +43     
+ Misses          618      600      -18     
+ Partials        574      570       -4     

Impacted Files	Coverage Δ
...tiny/operators/multi/MultiBufferWithTimeoutOp.java	74.78% <0.00%> (ø)
...lrye/mutiny/operators/multi/MultiOnItemInvoke.java	94.11% <0.00%> (ø)
...llrye/mutiny/operators/multi/MultiSkipFirstOp.java	90.90% <0.00%> (ø)
...allrye/mutiny/operators/multi/MultiSkipLastOp.java	82.75% <0.00%> (ø)
...rators/multi/overflow/MultiOnOverflowBufferOp.java	91.15% <0.00%> (ø)
...tors/multi/overflow/MultiOnOverflowKeepLastOp.java	91.30% <0.00%> (ø)
...lrye/mutiny/operators/multi/MultiSelectLastOp.java	87.50% <25.00%> (ø)
...llrye/mutiny/operators/multi/MultiCollectorOp.java	95.65% <50.00%> (ø)
...allrye/mutiny/operators/multi/MultiLastItemOp.java	88.46% <50.00%> (ø)
...tiny/operators/multi/MultiOnSubscribeInvokeOp.java	90.00% <50.00%> (ø)
... and 28 more

[jponge]

@cescoffier I just added a UnicastProcessor footprint reduction.

[jponge]

@cescoffier If you want to have a final look (but do not merge).
There are other places (including these classes) where we could reduce allocations, but that's already a good step and we don't have to track all of them.

@anavarr did not spot any performance hit, and he's looking at confirming allocation rates reductions.