setting initial log level to WARN, skips lots of initial log messages set at INFO level

The issue is that all of these messages are printed at INFO level and the default `loglvl level = LOGMSG_WARN;`. There is no chance to set the log level to INFO or lower (in the lrl) and be able to see the following initial messages -- the lrl options are parsed /too late/ for the following messages:
```
ulimit for data already set
adjust_ulimits: set ulimit for no. fds to 1048576
setting i/o alarm threshold to 100ms
__berkdb_write_alarm_ms = 100
__berkdb_read_alarm_ms = 100
__berkdb_fsync_alarm_ms = 100
thd_init: thread subsystem initialized
Plugin 'logdelete' installed.
Plugin 'logdelete' installed.
Plugin 'logdelete' installed.
Plugin 'newsql' installed.
Plugin 'remsql' installed.
Plugin 'remtran' installed.
Plugin 'alias' installed.
Plugin 'repopnewlrl' installed.
Plugin 'queuelua' installed.
Plugin 'queuedynlua' installed.
Plugin 'sockbplog' installed.
machine_class_default(titani.home) open rc -1 cdb2_get_dbhosts: no default_type entry in comdb2db config.!
registered resource lrl -> /dev/shm/adidb/adidb.lrl
Schema module init ok
Set default log level to DEBUG
```

The only way to see them is to set initial log level to INFO.
Maybe (some of) those messages should be raised to WARN or the initial log level should be set to INFO.
Or things are ok with the way they are, but I'm submitting this anyway for awareness.

No file or test is named. Start by tracing startup log initialization and the point where lrl options are parsed, then compare that order with the listed INFO messages. The issue does not define whether the intended result is changing message levels or changing the initial log level, so the completion criteria need maintainer agreement.

Written by the indexing model from the issue text.