Print more information in `show`

[nalimilan]

I suggest we give more information when printing models than just the formula and coefficient table. Changing the result of show may not be considered as breaking, but 2.0 would be a good time to do this. I would suggest printing at least these indicators which are present in R, Stata and SAS:

• number of observations (super important in particular to check that you didn't lose too many observations due to missing values)
• R² and adjusted R² (for linear models -- it's more controversial to choose a pseudo-R² for GLMs)
• log-likelihood (for GLMs)
• F-test or LR test
• dispersion parameters for distributions that have one

Other possible indicators to include:

• AIC and AICc
• null log-likelihood

Econometrics.jl seems like a good inspiration:

Continuous Response Model
Number of observations: 223174
Null Loglikelihood: -6746726.41
Loglikelihood: -345253.18
R-squared: 0.0466
Wald: 860.46 ∼ F(3, 223158) ⟹ Pr > F = 0.0000
Formula: log(Earnings) ~ 1 + Educatn + Age + (Age ^ 2) + absorb(Kids)
Variance Covariance Estimator: OIM
──────────────────────────────────────────────────────────────────────────────────────
                    PE           SE      t-value  Pr > |t|         2.50%        97.50%
──────────────────────────────────────────────────────────────────────────────────────
(Intercept)   8.23837      0.13135      62.721      <1e-99   7.98093       8.49581
Educatn       0.000538588  0.000138739   3.88203    0.0001   0.000266663   0.000810512
Age           0.0389003    0.0068679     5.66408    <1e-07   0.0254394     0.0523613
Age ^ 2      -0.000191192  8.86411e-5   -2.15692    0.0310  -0.000364926  -1.74575e-5
──────────────────────────────────────────────────────────────────────────────────────