UpgradeChecker: fix the stable version number

[melissalinkert]

Fixes https://trac.openmicroscopy.org.uk/ome/ticket/11865.

[ghost]

This looks fine and good to merge.

However, do we really need to embed a (soon outdated and static) version in the source? Can't we query this externally using the qa server, so the only information we need to hand is our own release version?

[melissalinkert]

The only problem I see with querying the registry is that it would need to return a different version depending upon what the current version is - i.e. the latest version is 4.4.10 if you're running 4.4.9, but it's 5.0.0-rc1 if you're running 5.0.0-beta1.

[joshmoore]

It's been discussed having both available, but no action has taken place yet (/cc @bpindelski)

[bpindelski]

@joshmoore The changes in this PR go along the lines with what already is on dev_4_4. Yes, I agree with @rleigh-dundee that ideally the QA server (or any server responding to the upgrade checks) should do some rudimentary logic to detect the major version of the requesting client. At the moment that would necessitate changes in the QA code. Trello ticket?

[melissalinkert]

Knowing that we need to put a pin in registry/QA changes, is there any reason not to merge this before the 5.0.0 code freeze?

[joshmoore]

Not from me.

[melissalinkert]

--rebased-to #872