Implement quite mode for alive web server

[andrcuns]

This PR implements suggestion from https://github.com/arturictus/sidekiq_alive/issues/85 to gracefully handle extended sidekiq shutdowns. Additional improvements and fixes are added as well.

• Implement "quite" mode when sidekiq is shutting down. While web server is in "quite" mode, it will return healthy status by default because alive key is not updated during this period
• Add a configurable timeout for quite mode so server is not stuck as healthy indefinitely if sidekiq process fails to shut down
• Attempt to fix timeout errors by not using sidekiq-alive capsule connections during the shutdown. Related to: https://github.com/arturictus/sidekiq_alive/issues/100
• Don't call callbacks and cleanup operations twice, purge jobs on quite, cleanup queues and call shutdown callback only on shutdown

Closes #85, #100, #105

[andrcuns]

@arturictus This is a follow up with some further improvements. I tested it out in my own k8s deployment and everything seems to be working as expected 🎉