Metrics API enhancement

[qqmyers]

Multiple metrics-related issues (such as #6766, #3313, #3527) envision per-dataset metrics. The existing metrics APIs (api/info/metrics and /admin/makeDataCount) do not yet support this and differ in other ways (e.g. how a time-series is reported, error reporting, published or draft datasets, whether objects or sql queries are used).

As part of a QDR effort to implement metrics reporting, I've started trying to standardize aspects of these APIs and, drawing on the code from DANS mentioned in #6766, to add per-dataverse capabilities. I'm submitting a draft PR to make the current state of this work visible to the community and am looking for any/all feedback as to how this can be adapted/extended to address related metrics issues and other community needs.

[qqmyers]

https://docs.google.com/spreadsheets/d/1MxcaTtK4Uq_7-4HGt-X2C6hYo7FzrSYYH2mt3ln4M3w/edit?usp=sharing has two sheets -

• the first lists the existing APIs related to metrics
• the second shows the additions/changes made to the APIs in this PR, with the changes/additions highlighted in yellow

The most significant changes include:

• /api/info/metrics/* endpoints can now be called per-dataverse - with results scoped to the tree of dataverses specified by a parent alias (with root as the default) (leverages the work from DANS - thanks!)
• most /api/info/metrics/* endpoints that return more than a single number now return either json or csv outputs
• there are timeseries outputs (so one doesn't have to make one call per month to assemble a time series)
• there is an endpoint listing the tree of subdataverses from a specified parent
• there are endpoints that report an MDC metric for all datasets within a specified parent dataverse (versus the existing mdc api being one dataset at a time)
• new endpoints to report file count and aggregate size per mimetype
• new 'uniquedownloads' endpoints that counts the number of unique downloaders for datasets (all downloads over time by one person, for one or more files in a dataset, all counts as one count. Intended to help in assessing which datasets are popular when datasets may have very different numbers of files.)
• general cleanup

I'm working now on changes to the dataverse-metrics app to allow use of these endpoints. Other than fixing any bugs I find, I'm not currently planning to do more work on the APIs themselves unless there are comments/requests/feedback on this PR. Depending on what that feedback is I may be able to make changes or we can add/update other issues.

[qqmyers]

FYI: The newmetrics branch of dataverse-metrics splits that app into two - a installation-level app showing many of the metrics in this PR (and allowing per-sub-Dataverse metrics), and the original global app that aggregates from any/all dataverses around the world. Still doing some testing/tweaking before making a PR(s) there, but any feedback welcome (from whether these two apps should really be in the same repo to look and feel, etc.).

[qqmyers]

Just added download counts by/per file id/pid and unique counts per file endpoints and made minor fixes (added csv for uniquedownloads.) Also added a file donwloads by id graph in dataverse-metrics.