Support USE statements, which may optionally have quoted identifiers

[bbeaudreault]

@alainjobart was having trouble finding tests for this particular area

[bbeaudreault]

@sougou thoughts on this? i don't think it's superseded by #2751

[sougou]

Forgot to mention. #2751 will be followed by a more universal implementation of USE that will make this obsolete, which is why I haven't approved this.
We were waiting for 2.1 to be cut. So, now I can submit that after @michael-berlin approves, and work on the subsequent PRs. :)

[bbeaudreault]

Ok sounds good, thanks. If that's the case I'm going to hijack this PR for the travis testing and add my unapproved code snippet from #2677 for inverting the single-db /0 logic. It will make it easier on my developers who are using the mysql prompt while we wait for the followup PRs.

[bbeaudreault]

@sougou where did we land with:

Forgot to mention. #2751 will be followed by a more universal implementation of USE that will make this obsolete, which is why I haven't approved this.

I know you added USE support, but do we still have cases where doing USE foo/0 results in more query flexibility than USE foo?

This branch is a pain to keep up-to-date as I pull in upstream changes. But we're starting to roll out vitess to more people now and I don't want them all to have to start remembering to always add a /0 when running ad hoc queries with the mysql server plugin

[sougou]

For all practical purposes. foo and foo/0 should work the same. There may be some exotic corner cases where it may not work. So, this can be closed.

[bbeaudreault]

Ok I'll close this for now and enter a separate issue if we run into issues. Thanks