Review how quantiles are scanned

[perrette]

Currently the fitting is done by searching for specific variable names in rimeX.records.fit_records (commit 33b0aa0)

This was a quick hack, but clearly it should be done in a more general manner, especially for the preferred case when a quantile or percentile column is present in the data. In the latter case, the variable name need to be stripped from the percentile mention before grouping. A user-define parameter may also indicate a custom mapping.

This is related and a pre-requisite to solving issue #16.