Expand support for Numpy NDArray functions

[Casper-Guo]

143 implemented Numpy NDArray support, opening the door to supporting a wide range of Numpy functions. Here is a list of frequently used functions and how they might be represented.

Some functions have unambiguous mathematical notations:

• [x] np.transpose: $a^T$
• [ ] np.linalg.vdot: $a \cdot b$
• [ ] The behavior of np.dot is more general and may requires some special handling
• [ ] np.linalg.inner: $\langle a, b \rangle$
• [ ] np.linalg.matmul: $ab$
• [x] np.linalg.matrix_power: $a^n$
• [ ] np.linalg.kron: $a \bigotimes b$
• [x] np.linalg.qr: $QR(a)$
• [x] np.linalg.svd: $SVD(a)$
• [x] np.linalg.det: $det(a)$
• [x] np.linalg.matrix_rank: $rank(a)$
• [x] np.linalg.inv: $a^{-1}$
• Although perhaps not the most rigorous, it should be acceptable to represent np.linalg.pinv as $a^{-1}$

Some other functions have standard representations in some cases (for example, when all the keyword arguments are set to the default). Refer to Numpy documentation, especially the calculations section, for many such functions. Note that some of these functions are implemented both in numpy and np.ndarray.

[ZibingZhang]

This feels like a bunch of function expansions.

[odashi]

some of them can be implemented by the existing tools. For $a^T$, it shouldn't be implemented as pow(a, T) because it is not actually the power operator.