Support Trace Schema for Reading

[DavePearce]

The goal here is (roughly speaking) allow reading trace data in conjunction with schema information. In particular, to maxmise space reductions for bytes and bit columns. That is, when reading a trace column with a width of 2 bytes but a schema type of :u8, the column data is automatically truncated accordingly. This should be particularly useful for bit columns.

[DavePearce]

This is less of an issue since many aspects were fixed at the source. However, for columns which take less than a u8 then it still applies. And there are a lot of these.

[DavePearce]

Roughly, then the plan would be:

• (Schema) A schema generates an initially empty trace, but with columns all the right size.
• (Reader) A reader populates the trace with available trace data. That should include the input columns, but could also include all trace data.
• (Expansion) At this point we can perform expansion.

Potentially, this process could all be wrapped up into some notion of trace builder. The building process can fail in various ways, such as if there are missing input columns or duplicates, etc. Also, if a field element doesn't match the input field (depending upon the mode of operation).