[Build Log classifier] Long log files give error when sent for classification to the Seldon service

[Shreyanand]

Describe the bug
When we send a request to the service at this point http://build-log-clustering-ds-ml-workflows-ws.apps.smaug.na.operate-first.cloud/predict, we see a 502 error when the log file is huge (>100,000 chars).

To Reproduce
Steps to reproduce the behavior:

1. Make sure you have the local cache of data by running the build_log_EDA notebook.
2. Go to model_seldon.ipynb
3. Select a log file with length greater than 100,000
4. Send it to the service endpoint using the notebook
5. See Remote disconnected or Error 502 from server

Expected behavior
The log file predictions should be available for all log files with server status 200.

Additional context
On inspecting the pod logs, it seems the service can't find json in these cases. I think it is either because the length of the log files makes them difficult to send through the post request or because these files are big, they have a higher chance of having json parse errors.