feat: "Aegis CI"

[bctnry]

currently we do this by doing the following:

keeping "one single executable" may not be the best way forward;

1. while saving repo config, an update hook is inserted into the actual git repo which triggers aegis ci with necessary info (e.g. with all the info aegis web-hooks would receive)
2. the config for aegis web server contains the ci server's endpoint address. this would be the address used by aegis ci or inserted into the hook.
3. when commits are pushed to the instance, the update hook would trigger, and the command aegis ci would be executed.
4. aegis ci would push a message to the ci server's queue.
5. aegis ci server runs separately from the web server. they may share the same resources (e.g. databases) with the web server, but the ci server is largely independent. it takes messages from the queue and processes them accordingly.

[bctnry]

need a place to config secrets and variables...