respect log level for all sources

[ilyam8]

Summary

There are two things in the logging library that are not documented and I believe are not bugs, but are implemented for a reason:

1. The default log level affects only the daemon and collector logs (nd_log_set_priority_level()).
2. There is a way to set the log level per source, but again, the filtering affects only the daemon and collector logs (netdata_logger()).

I think I understand 1. - we don't want to filter health (alerts transitions) and access sources (web server access logs) if the global log level is < info. But I think we should respect the log levels of each source when explicitly set (2.).

The issue is reported on Discord.

This PR changes both 1 and 2: both global and individual log levels apply to all sources.

Test Plan

Set access log level, check logs.

Additional Information

For users: How does this change affect me?

[ilyam8]

@stelfrag @ktsaou Do you think that health, access, and aclk are special sources and we need to treat them differently? Like they only can be enabled or disabled, and log-level filtering is not applicable for them?

If so, I will close this PR and just update the documentation. The only problem is that these special sources contain debug-level messages which can not be filtered.

[thiagoftsm]

A quick update, while we are waiting some answers. On Slackware and FreeBSD logs are working as expected.

[thiagoftsm]

I am considering the @stelfrag approve, the answer we were expecting, so LGTM!