Be sure to rebuild the generated documentation with the proper version string

[axch]

ditto bdbcontrib.

I believe there are two non-automatic things about this now:

• Updating the reference manual's idea of the current version
• Actually rerunning the documentation build and uploading it to the canonical documentation location

[riastradh-probcomp]

What properties characterize the resolution you would hope to attain for this bug?

The reference manual's version should be bumped only when we actually know it documents the newer version -- it is intentional that the build does not automatically grab bayeslite/VERSION or anything.

Automating publication of the documentation sounds like it is a subproblem of automatic how we publish everything -- and we don't have a single process in place for that at the moment. Did you have anything more specific in mind about that concerning documentation?

[axch]

I suppose there are three things I want, which may plausibly deserve to be different bugs:

• Ensure that the reference manual published with R5 both documents the R5 software and claims to.
• Define an automatic release process that includes automatic publication of reference documentation
• Satisfy ourselves about the processes (automatic or manual) whereby we maintain the freshness of our reference documentation.

[riastradh-probcomp]

I updated the version in doc/conf.py for 0.1.3.

[riastradh-probcomp]

commit 257856d

[gregory-marton]

@raxraxraxraxrax, you may want to review and recommend changes for the documentation for 0.1.4, taking into account whatever does actually happen for polish, and I guess keeping in mind the question of how in particular satellite enthusiasts might interact with it, if at all. It would be good to have some part of the release process checklist that includes such a review for the future, rather than relying on a bug?