Add script to compute Anomalies in full cross-validation

Hi @nperez, here is the MR to add Crossval_anomalies.R to master.

Before merging, I have two things I wanted to ask:

1. I compared the results of this multiApply version to the results of the original version. Everything is the same except for the forecast limits, which were a little different. I realized it's because in the original version, they are computed after rounding the thresholds to two decimal points (https://earth.bsc.es/gitlab/es/sunset/-/blob/dev_cross/modules/Crossval/Crossval_anomalies.R#L150) while the limits for hindcast and obs are calculated before rounding. In this new version, all the limits are calculated before rounding. Is that okay?
2. I wanted to add Crossval_metrics.R as well, but I have noticed that for the RPS and RPSS it may not work as expected when more than one set of percentiles is requested, because the loop over categories rewrites the results as $rps/$rpss each time. So I think it may be better to wait until we have discussed how we want to approach this case, and fixed this. Let me know what you think.

P.S. I think the pipeline is failing because esarchive is not mounted on the gitlab runners yet, I have run it locally and the tests pass.

Victòria