Proposal

Proposal:
Introduce a new PromQL function, histogram_quantiles(b instant-vector, φ0 scalar, ...), as a variadic implementation of histogram_quantile, capable of computing and returning multiple quantiles in one query. The results for each quantile would be differentiated by a synthetic label (say, __hist_quantile__), the value of which would be a value of quantile argument which was used to compute the given result. The function would work with both classic and native histograms.

For example:

histogram_quantiles(single_histogram, 0, 0.5, 0.9, 1)
    {__hist_quantile__="0"}   0.5
    {__hist_quantile__="0.5"} 1.414213562373095
    {__hist_quantile__="0.9"} 3.0314331330207964
    {__hist_quantile__="1"}   4

Motivation:
Dashboards that are rendering multiple quantiles on a single panel. Currently, the user experience suffers, because the singular histogram_quantile query (sometimes with a really complicated argument) has to be repeated (and later maintained) separately for each quantile (have seen 5 or even more in practice).
Additionally, having this result available from a single query would improve dashboard performance and reduce query load on the system.

Alternatives:
Arguably, this way of rendering a histogram somewhat competes with heatmap graphs, but for many use cases, plotting concrete quantiles is more precise, discrete and easier to consume:

Prior art:
There was a similar proposal discussed in 2016: #1547
However, it seems it was targeting a different use case (recording rules), besides, 9 years have passed since then.

Implementation:
The function would reuse funcHistogramQuantile called for each value of quantile argument, in a loop.
Can help with the implementation in case this feature request gets accepted.