Retry operations with distribution repositories, such as `apt update`

[Nick Volynkin]

apt update or yum update in package jobs sometimes fail due to repository being temporarily unavailable. It's a normal behaviour for a repository and network, we cannot expect 100% reliability from them. So, we should learn to overcome this problem.

One of possible solutions is to retry all repository operations in the build script. Making an extra attempt or even ten extra attempts is still much better than restarting the job manually.

I'd propose that we retry update tasks 10 times with 10 second intervals. Both numbers are subject to discussion.