output/cloudv2: Flush chunks

[codebien]

If the number of time series is higher than the maximum batch size than split them in chunks.

Part of #3117

[codecov-commenter]

Codecov Report

Merging #3108 (7fd710a) into master (0a31512) will increase coverage by 0.07%.
The diff coverage is 55.55%.

❗ Current head 7fd710a differs from pull request most recent head 6e63afd. Consider uploading reports for the commit 6e63afd to get more accurate results

@@            Coverage Diff             @@
##           master    #3108      +/-   ##
==========================================
+ Coverage   73.74%   73.82%   +0.07%     
==========================================
  Files         241      243       +2     
  Lines       18460    18474      +14     
==========================================
+ Hits        13614    13638      +24     
+ Misses       3972     3966       -6     
+ Partials      874      870       -4     

Flag	Coverage Δ
ubuntu	73.74% <55.55%> (-0.01%)	⬇️
windows	73.64% <55.55%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
cloudapi/config.go	89.47% <ø> (ø)
output/cloud/expv2/flush.go	84.61% <50.00%> (+0.74%)	⬆️
output/cloud/expv2/output.go	84.33% <100.00%> (+0.19%)	⬆️

... and 11 files with indirect coverage changes

[mstoykov]

I was bit weary of an if in this loop ... but to be honest addTimeBucket is so much more complicated that this is likely not an issue.

If it is we can always optimize it.

[mstoykov]

LGTM!

Left a nitpick that is unrelated to the current PR

[imiric]

I still don't fully grok the new cloud output, but this change LGTM.