Add an option to keep retrying when MariaDB encounters a full disk error while trying to open a new binary log file

MariaDB already retries writes when full disk situation is encountered while writing to the existing binary log file. See https://mariadb.com/kb/en/library/using-and-maintaining-the-binary-log/#effects-of-full-disk-errors-on-binary-logging:

"If MariaDB encounters a full disk error while trying to write to a binary log file, then it will keep retrying the write every 60 seconds. Log messages will get written to the error log every 600 seconds."

This is a request to add an option (maybe set by default) to get the same behavior for the opening new binary log file case, instead of current:

"...if MariaDB encounters a full disk error while trying to open a new binary log file, then it will disable binary logging entirely."

Disabling binary logging prevents further point in time recovery even after disk space is freed and logging is again enabled.

See also related MDEV-17856. When that feature is implemented and IGNORE_ERROR is set, retries should continue for opening new binary log as well.