Regression problems

[Michael-Howes]

Add regression problems with Capital Bikeshare dataset https://capitalbikeshare.com/system-data

Description of benchmark problems:

• Large regression problems with ~1_000_000 observations and ~300 features.
• Linear regression problems with ridge, lasso and elastic net regression (Clarabel solve time ~15 s).
• Logistic regression problem with lasso regression (Clarabel solve time ~150 s).

New files:

• Add problems/ecp.py for exponential cone problems (logistic regression).
• Add data/bikeshare.py for loading data.

New dependencies:

• Add scipy, pandas and requests.

[SteveDiamond]

@PTNobel what's your view on the benchmarks downloading external data? Probably better to include the data as a git lfs file IMO.