Fix global status metric names in mysqld_exporter

User impact: User doesn't see MySQL metrics
Steps to reproduce:

1. Install MySQL 8.0 and install "validate password" component
2. Create user for PMM
3. Add MySQL to PMM

Actual result:
pmm-admin check-network returns that mysql:metrics service is DOWN and there are errors in prometheus.log:

[[email protected] opt]# tail -50 /var/log/prometheus.log 
level=warn ts=2019-01-29T09:08:03.050992552Z caller=scrape.go:713 component="scrape manager" scrape_pool=mysql-hr target=https://10.10.11.53:42002/metrics-hr msg="append failed" err="no token found"

Expected result: All services are running

Original report:
For Mysql 8.012  with  https://dev.mysql.com/doc/refman/8.0/en/validate-password-options-variables.html#sysvar_validate_password.dictionary_file 

Mysqld_exporter reports metrics with "." inside  name and this is invalid for Prometheus so it skips  the metrics and writes a message to the logs  

based on : 

https://github.com/prometheus/mysqld_exporter/pull/337 

and 

https://github.com/prometheus/mysqld_exporter/issues/325#issuecomment-426798030?

and 

https://www.percona.com/forums/questions-discussions/percona-monitoring-and-management/53180-pmm-mysql-8-0-11-instances-not-showing-graphs/page2

TODO: port changes from upstream to temporary  fix the problem before we\ll update exporter