Investigate speed of Critical Hours calculation

**Describe the task**
The current algorithm for calculating Critical Hours is at risk of being considered slllloooooowwwww because it's just brute-forcing. With a smarter algorithm we could make this calculation much more efficient and get results much more quickly.

**Acceptance Criteria**
- [ ] Research how long it takes to calculate critical hours for (a) 1 station, (b) many stations (with "many" being arbitrarily defined - perhaps 20-30?)
- once this has been investigated, consult with team & PO - is our current solution quick enough? If not, proceed to next AC
- [ ] It should take time on the order of milliseconds to calculate 1 station's critical hours, and it should take less than 1 second to calculate the critical hours for "many" (20? 30? more?) stations (in 1 request). If time exceeds 1 second, then write a ticket to make it faster

**Additional context**
- possibly use Newton's method (algo is implemented in scikit), apply it to vector of values using pandas' `apply()`

Locate the existing Critical Hours calculation entry point and benchmark one station versus many stations, defining “many” as part of the investigation. Compare the results with the millisecond and one-second acceptance criteria, then consult the team and PO; if the limit is exceeded, record a follow-up ticket for optimization.

Written by the indexing model from the issue text.