Less Fragile Report Generation

[coreyostrove]

There are a number of places in the report generation that are more fragile than they should be. Examples include:

Issue #278
Errors related to clifford compilations for custom gate sets
Etc. (other examples are escaping me, others should chime in here).

Ideally we should be more robust to these failures, sticking the generation of most plots in try except blocks and raising warnings when specific parts of the report generation fail rather than allowing the entire report generation to fail. We could then simply exclude the plots/tables that couldn't be generated with a message about why.

[sserita]

Generally maybe have figures/tables always generate errors but we catch and keep going, i.e. HTML philosophy.

[coreyostrove]

Note to selves, some of the changes in PR #615 and PR #572 constitute partial progress toward this (hazily defined) goal.