Allow for BMI and height to be ranges

[wleoncio]

This should also make calculation of the confidence intervals (#9) easier.

Test with one example (simple FX, vo2_ml_kg_min):

• Group $\beta$s in matrices, lists, whatever works best (1 per group)
• Calculate weighted.mean() without relying on apply().
• Calculate $y$ as a function of the new beta and x
• Make sure new output match the ones from v1.0.0.9001
• Allow $X$ to (internally) be a $k \times p$ matrix, where $p$ is the number of parameters and $k$ is the number of observations
• Allow input to be a range of $k$ observations for height and BMI
• Make sure new output match the ones from v1.0.0.9001

Generalize to all others:

• Group $\beta$s in matrices, lists, whatever works best (1 per group)
• Calculate weighted.mean() without relying on apply().
• Calculate $y$ as a function of the new beta and x
• Make sure new output match the ones from v1.0.0.9001
• Allow $X$ to be a $k \times p$ matrix, where $p$ is the number of parameters and $k$ is the number of observations
• Allow input to be a range of $k$ observations for height and BMI
• Make sure new output match the ones from v1.0.0.9001

[wleoncio]

Reference values from v0.0.0.9002:

[wleoncio]

See #4 (comment) for tips on object sizes.