Update point estimate at alpha == 1 docstrings

[stefan-apollo]

rib % grep "point estimate" . -r --include="*.py" --color
./rib/linalg.py:            point estimate at alpha=1 instead of using the trapezoidal rule.
./rib/linalg.py:            point estimate at alpha=1 instead of using the trapezoidal rule.
./rib/hook_fns.py:            to taking a point estimate at alpha == 1.
./rib/hook_fns.py:    integrated gradient effectively takes a point estimate for the integral at alpha == 1.
./rib/hook_fns.py:            to taking a point estimate at alpha == 1.
./experiments/lm_rib_build/run_lm_rib_build.py:        "If 0, we take a point estimate (i.e. just alpha=1).",
./tests/test_linalg.py:    The point estimate should be double the value of n_intervals==1, and n_intervals==5 given that