In fact I have just completed a few backports and non field db data into reports (Min Order, Prepaid Freight, etc) in my FAMods besides the en_GB-general.sql complete fix in the extensions.
The devs have categorically said they will only provide critical updates to the v2.3.x branch henceforth and I expect a final v2.3 release is hence due.
On the 7th the extensions were queried and an update was expected but all were frozen as of 28th last. I have also done a thorough clean up of all 38 official languages including the yet unaccepted Tamil (ta_IN) even though it has more phrases translated than some of those already there.
The easiest way to do a fresh CoA is to
1. Import the en_US-new.sql into a fresh database
2. Export it from your mysql client schema + data - to say one.sql
3. Make changes as you see fit in the tables that contain data - Currencies, date/time formats, company defaults, sys_prefs, chart_masters, fiscal years, etc.
4. Export it to say two.sql.
5. Use Winmerge to sift the differences and see how they compare with the existing Canadian one.
6. Make changes to two.sql but retain a blank audit_trail table and the standard users table from the US CoA except for some specifics for Canada. That way you'll preserve the default admin access / password and not bother with session based timestamp changes.
7. Remove the CREATE DB and USE DB commands in two.sql.
8. Now you have your CoA - two.sql - rename it to en_CA-general.sql and upload it in the place of your officially installed one so that the config entries remain.
The ones I maintain have the schema without the collations bunched together and the insert statements entered later all in one bunch with those that vary between CoA's alone towards the end. That way when making a new CoA, we just need to copy this and make changes to the data towards the end alone!
In fact my suggestion was to keep the schema and standard data in one sql file and the rest that changes between CoAs alone be used from separate sql files. It remains unaccepted due possibly to accommodate some queer collations and out of turn insertion of data where strict InnoDB and other RDMS constructs will make import fail on cascade dependencies.