Document semantics of TimeSpans.index_from_time(sample_rate, span::TimeSpan)

[ericphanson]

The current implementation of index_from_time(sample_rate, span::TimeSpan) makes an undocumented choice of semantics, which is to round-down the endpoints, and ensure the right endpoint of the span is excluded.

What do I mean this is undocumented? Well,

integer index of the most recent sample taken at sample_time

documents the index_from_time(sample_rate, time) method. That isn't sufficient to learn the semantics of the span method, since we do something other than simply call index_from_time on each endpoint, re

TimeSpans.jl/src/TimeSpans.jl

Line 237 in eb8b794

j = i == j ? j : (j - 1)

and we only document the span method as

TimeSpans.jl/src/TimeSpans.jl

Line 221 in eb8b794

Return the `UnitRange` of indices corresponding to `span` given `sample_rate` in Hz:

which leaves it unclear exactly what UnitRange we're talking about.

Originally posted by @ericphanson in #26 (comment)