Note that HERE %iles are interpolated values, not necessarily observed in samples

[Nate-Wessel]

I.e. HERE calculates percentiles with something like PostgreSQL's percentile_cont rather than percentile_disc as I had originally assumed. This means that except where the sample_size = 3 (for the median) or sample_size = 20 ( for the 85th), the percentile values are almost certain to be interpolated rather than actually observed in the sample.

This can be demonstrated by looking at records WHERE sample_size = 2

SELECT 
    min_spd,
    pct_50,
    max_spd
FROM here.ta_path
WHERE sample_size = 2
LIMIT 100

What this pull request accomplishes:

• Notes a quirk in the HERE data

Issue(s) this solves:

• adds to documentation

What, in particular, needs to reviewed:

• facts

What needs to be done by a sysadmin after this PR is merged

NA