improvement: add projections to wt dump utility

I'd like to do something like this:
wt dump -j 'table:employee(id,department)'

when I try it I get
wt: table:employee(id,department): No such object exists.

A simple use case (especially with -j, or JSON output) is to edit out some fields. In these cases, one could use a JSON stream editor, like jq to edit the output. Another use (without -j) might be to reorganize a table, where it is reloaded into a new schema. Again, with JSON output and jq, one could consider doing this (using wt load -j). But another use case that cannot be solved with JSON stream editors is when dumping an index, and wishing to also see the primary key. For example,
wt dump -j 'index:employee:department(id,salary)'

could be a way to see the primary key "id" ordered by the index.

One consideration would be to adjust the metadata output for the table/index to indicate the new values of columns and value_format.