Change 'too many labels' error message to play nicely with Prometheus logs

[pracucci]

What this PR does:
Prometheus truncates remote write errors displayed in logs at 512 bytes (was increased from 256 to 512 in this PR). An user privately reported me that wasn't understanding why some series were rejected and Prometheus was logging the following:

ts=2021-01-19T14:30:48.352Z caller=dedupe.go:112 component=remote level=error remote_name=45f6d9 url=https://prometheus-us-central1.grafana.net/api/prom/push msg="non-recoverable error" count=99 err="server returned HTTP status 400 Bad Request: sample for 'istio_response_bytes_bucket{app_kubernetes_io_instance=\"infrabin\", app_kubernetes_io_name=\"infrabin\", connection_security_policy=\"unknown\", destination_app=\"unknown\", destination_canonical_revision=\"latest\", destination_canonical_service=\"infra"

Since the log is truncated, it's not clear the problem is "too many labels". In this PR I'm proposing to change the format of the "too many labels" error message to move the series at the end.

I took a quick look at other validation errors and I think this the only one suffering this issue.

Which issue(s) this PR fixes:
N/A

Checklist

• Tests updated
• Documentation added
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX]

[pstibrany]

👍

[bboreham]

Related: #1582

[bboreham]

lgtm

[pracucci]

Related: #1582

I'm very bad with the GitHub search apparently. I was looking for that issue and I haven't found it 😞