batch logging should watch the resulting batch payload size

[rplevka]

Describe the bug
We currently only issue the batch logging after the configured batch_log_size is about to be exceeded.
https://github.com/reportportal/client-Python/blob/master/reportportal_client/core/log_manager.py#L72

This method also needs to watch the current batch payload size, as RP server refuses requests that exceed the pre-configured maximum payload size:

ResponseError(error_messages[0])
16:10:24  reportportal_client.errors.ResponseError: 5000: Unclassified error [Maximum upload size of 67108864 bytes exceeded; nested exception is org.apache.commons.fileupload.FileUploadBase$FileSizeLimitExceededException: The field json_request_part exceeds its maximum permitted size of 67108864 bytes.]

Steps to Reproduce
Steps to reproduce the behavior:

1. either set the log_batch_size to a really high value and try to log a lot of messages or try to log a really long message (tl;dr generate a payload larger than a max allowed payload size on a server - default is probably 67108864 bytes)

Expected behavior

• implement configurable parameter: log_batch_payload_size.

• _log_process function should watch this and "flush" (send) the batch before appending another log message in case the payload is about to be exceeded.

• Figure out a sane way of handling a single log that itself exceeds the limit - maybe some string-splitting strategy could apply with a WARNING message being logged.

Actual behavior
the client crashes with raising the ResponseError

Package versions
5.2.0

[HardNorth]

Should be implemented here: https://github.com/reportportal/client-Python/releases/tag/5.2.3