motion: clarify "latency" channel type description

[sappelhoff]

Regarding the comment from @sappelhoff about latency channel, yes, the onset is intended to be the acq_time of scans.tsv and the text could be updated to make that clear.
That would also imply that the first sample should always have latency of 0, considering acq_time = when the first data point in each run was acquired.

ss[.000000] was not intended to imply integer digit limit. But I see how it is misleading.
Should we update the notation to s[.000000] and add that s refers to an integer of arbitrary number of digits?

Originally posted by @sjeung in #1436 (comment)

This issue is to explicitly track these points which came up in another thread. Just to prevent this from being buried and forgotten, as it's important.

Related to:

• [ENH] Allow fractional seconds in StartTime #1459

[JuliusWelzel]

See my solution at #1464. Is there a different file where the latency channel description should be edited, for example in the schema?