Add function(s) to create timestamp-to-timepoint mapping

# :rocket:

- [ ] Is your feature request essential for your project?
Not essential for `v0.1`, since we can start with an externally-defined mapping, but will be a high priority for `v0.2` to minimize the amount of information that we expect the user to provide in order to start a simulation.

### Describe the workflow you want to enable
I wish we had one or more algorithms which would take information about profiles (and a Grid?) and produce a mapping which will feed #55 to develop Switch inputs and feed #50 to analyze Switch outputs.

### Describe your proposed implementation
Algorithm TBD. We could start with something very simple like k-means in order to test the full chain of processing inputs to outputs, and eventually get a little fancier by doing e.g. some transformation to the profiles before generating clusters (e.g. principal component analysis).

The function would take a set of profiles, a number of desired timepoints, and possibly a Grid object, and produce a pandas Series which maps each timestamp of the original profiles to a timepoint (possible producing any additional information as necessary to apply the mapping in #55).