Issues when running evaluation over gold standard #6

As a sanity test on evaluating agents on DA-Code I checked the evaluation of the gold standard against itself. See https://github.com/michaelrglass/da-code/blob/gold-vs-gold/tests/self_eval_check.py

This revealed a few issues:

Easily fixed

• In calculate_list, the empty list is not equal to itself.
  https://github.com/michaelrglass/da-code/blob/270c59b36c7c961c82a4d4c73b83e2932cd52638/da_agent/evaluators/metrics/text.py#L20
  This is a problem in di-text-029 and di-text-035

• The file gold/ml-cluster-003/cluster.csv is not present, instead it is clustering.csv
  However, renaming this file gives a total score of 0.005763917074549833. Like a number of other 'ml' instances this has a non-zero, non-perfect score.

Unknown failures

• data-wrangling-038 (maybe because it is a .db file?)

ML score below 1

• ml-cluster-001
• ml-cluster-002
• ml-cluster-003
• ml-cluster-004
• ml-cluster-006
• ml-cluster-007
• ml-cluster-008
• ml-cluster-009
• ml-cluster-010
• ml-cluster-015
• ml-cluster-017
• ml-cluster-018
• ml-cluster-021
• ml-competition-003
• ml-competition-005
• ml-competition-007
• ml-competition-009
• ml-competition-010
• ml-competition-011
• ml-competition-013
• ml-competition-014
• ml-competition-015
• ml-competition-019