Based on the measured data set the user/namespace/project limits

[lbarcziova]

To avoid overusing Packit, we will implement a system of limits. Here are the rules:

• Let's start with the hourly base for the limits.
• Specific limit numbers should be determined from real usage.
• Be prepared for manual enhancements of limits (e.g. a known important project with high traffic that uses the resources legitimately).
• Be prepared for integrating with Implement blocklisting mechanism #1964 (later)
• If the limit is reached, no action is done by Packit.
• Can we inform user about this situation so it's realised by a legitimate user? (One-time action, two levels of limits,...)

Followup of #1990