Fix Shannon Entropy: sum instead of mean

[knife982000]

This PR addresses a bug in the computation of the Shannon Entropy metric used in the recommender system.

🐛 Bug
The original implementation incorrectly computed Shannon Entropy as the negative mean of p(x) * log(p(x)), which deviates from the standard definition of Shannon Entropy:

H(X) = -∑ p(x) * log(p(x))

✅ Fix

• Replaced the mean with sum in the Shannon Entropy calculation to correctly reflect the definition.

🧪 Tests

• Existing test for Shannon Entropy has been adjusted to align with the corrected implementation.
• Changed assertEqual to assertAlmostEqual with a delta of 1e-15 to account for minor discrepancies due to floating point precision.
• Test continues to pass with a 1e-15 delta, confirming numerical accuracy.