[BUG] Random Forest fails to generate output with date in the past

[verovaleros]

When AIP is run on an earlier date, all models succeed except RF:
docker run --rm -v /opt/zeek/logs/:/home/aip/AIP/data/raw:ro -v ${PWD}/data/:/home/aip/AIP/data/:rw --name aip aip:latest bin/aip --date 2024-10-27 -v

2024-10-29 08:35:49,913 - root - ERROR - Error running Random_Forest model: date value out of range
Traceback (most recent call last):
  File "/home/aip/AIP/bin/aip", line 56, in run_model
    blocklist = aip_model.run(date_day)
                ^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/aip/AIP/lib/aip/models/prioritize.py", line 255, in run
    training, target = self._get_training_target_set(for_date=for_date)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/aip/AIP/lib/aip/models/prioritize.py", line 240, in _get_training_target_set
    training = Knowledgebase(load_until=for_date - timedelta(days=2))
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/aip/AIP/lib/aip/models/prioritize.py", line 116, in __init__
    self._load_knowledge_until(day)
  File "/home/aip/AIP/lib/aip/models/prioritize.py", line 106, in _load_knowledge_until
    self.build(end=datetime.strptime(day, '%Y-%m-%d').date())
  File "/home/aip/AIP/lib/aip/models/prioritize.py", line 133, in build
    day = str(end - timedelta(days=days_ago))
              ~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~
OverflowError: date value out of range