ENH: optimize: add _chandrupatla_minimize, a vectorized scalar minimizer

[mdhaber]

An elementwise implementation of Chandrupatla's scalar minimization algorithm. Uses the framework developed for Chandrupatla's scalar root-finding algorithm.

---

Original post

@tupui here is a draft of _chandrupatla_minimize that needs a bit of feedback before it can go further. The main question is how you want this to be factored. To best understand the question, it helps to understand what I did here:

• The first commit copies _chandrupatla into a new _chandrupatla.py and TestChandrupatla into test_chandrupatla.py.
• The second commit simply swaps some imports in the tests so that test_chandrupatla.TestChandrupatla is testing _chandrupatla._chandrupatla instead of _zeros_py._chandrupatla.
• The third commit converts _chandrupatla._chandrupatla into _chandrupatla._chandrupatla_minimize and test_chandrupatla.TestChandrupatla into test_chandrupatla.TestChandrupatlaMinimize.

The point is that by looking at the third commit, you can see what is shared between the root finder and optimizer.

With that in mind, how do you want to see this? What file(s) should things be in, and what code do you want to share between the two?