Fix miscellaneous issues with deploy_heroku.py

[CreatedBySeb]

1. The suffixes Heroku began adding to app names were not being correctly handled by the script, I've modified the RegExp to remove them and tested that this works correctly
   • This would otherwise cause a crash when the script attempted to add the Heroku Redis addon due to the script attempting to include the suffix when specifying the app
2. There has been an intermittent issue with deployments failing via web deploy or via command line because Heroku Redis was taking too long to provision, so by the time we reached the deploy step it was not ready and the deploy would fail due to the missing REDIS_URL environment variable
   • This issue seems to mostly be resolved on Heroku's end, however there are never any guarantees Redis is provisioned by the time we attempt to deploy Tabbycat, and the time taken to reach this step could depend on numerous factors including future changes, so it is still worth fixing
   • This fix essentially polls the Heroku CLI every 5 seconds to see if Redis has finished provisioning, then proceeds, meaning a maximum unnecessary delay of 5 seconds, but better consistency
3. Currently the dyno type defaults to Eco in the script, which would cause a failure if the deployer did not have an Eco plan and forgot to manually specify basic dynos, instead by not specifying an explicit type as the default, Heroku will automatically use the Eco dyno if the user has an Eco plan, or a Basic dyno otherwise
   • The docs have been updated to reflect this behaviour and better explain how payment works now, since it is no longer really possible to deploy a 'free' Tabbycat site, unless you're using the student credits
4. There are a couple of different places where information was out of date relative to the Heroku changes regarding free services and other platform changes, so I updated these parts of the docs, including the app name suffix and Postgres changes