Hi Simon,
There is simply too much to cover here in order for a good understanding looking at your starting point. It requires a base knowledge of statistics and histogramming as well as experience at actually debugging query processing on ASE.
My advice is to start running some queries ( with statement cache off) using set option show_lio_costing on and set statistics plancost. There is no clear documentation that will explain everything you see but the only way to understand it is to work with it day in and day out.
Once you start to make sense of the individual lio costing ( you'll need to look at optdiag output to) you can expand up to set option show_search_engine for join evaluation then finally set option show long for a deep dive. If you want to go further you then have to start looking at show_histograms and so it goes on.
Start simple and look at single search predicate queries and expand up. Create a table, put a row in it, create an index, add another row different values, update stats, and so on and so on. Look at the data available at each iteration.
There is plenty of unofficial documentation around if you look for it ( even on his site).
The figures you mention are histogram frequency cells ( see optdiag), holding essentially a weight which is a proportion of a domain which matches a given value.
Cheers,
Simon