More robust match predictions

Match predictions are currently just the sum of the estimated contribution of each team on the alliance. This doesn't deal with some scenarios, and also doesn't give any information on the variance of the statistics. We can use the available data to make better predictions of which alliance will win a match, and determine a measure of how confident we are in our prediction.

See the following TODOs from the predict_match_score function:
```python
# TODO: Deal with the fact that only one robot can climb onto HAB 3
# TODO: Consider that only two teams can start on HAB 2
# TODO: Use variance to create a confidence interval of which alliance will win the match
```

Start at the predict_match_score function and review the three listed TODOs: HAB 3 climbing limits, HAB 2 starting limits, and variance-based confidence intervals. Use the available match data to address those scenarios and produce a confidence measure alongside the alliance prediction; the issue does not name a file or tests to run.

Written by the indexing model from the issue text.