Update IPC code to remove Cython

[randyzwitch]

Goal of this PR is to remove the Cython build process completely, which should improve compatibility across package install tools.

WIP as I'm still trying to figure out if we can release shared memory; original attempt to call shmdt during load_buffer() killed the kernel, suggesting that pyarrow is still using that memory at least up to _load_schema if not all the way through _load_data().

Consider passing ptr value out of load_buffer() function, then calling shmdt twice after data has been made into pandas dataframe (once for each original handle passed from Thrift)

[andrewseidl]

Haven't looked too deeply at the new code yet, but CPU-only select_ipc finally works for me with a simple pip install, no conda needed at all. Tested with 3.7 (after tweaking pyarrow to a supported version) and 2.7.

Thanks!

[randyzwitch]

Awesome @andrewseidl! Should be ready for a full review tomorrow.

[randyzwitch]

When this passes CI, it's ready for review.

Looking for someone who understands shared memory better than I do to validate that shmdt conceptually is doing what I expect, which is to detach the shared memory from the Python process. My understanding is that shmat brings the shared memory into the Python namespace, and my expectation is that calling shmdt, and calling the deallocate_* methods from Thrift will release the memory from Python and release the reference to it from OmniSci.

I call both of these memory-freeing methods after a pandas dataframe is created, so that the pandas df result will then be managed by the Python garbage collector, and no other IPC memory references should exist.

Using the Python debugger, shmdt returns the proper code from C for the given pointer, so the code is technically correct, I just need conceptual validation.