Simplify building process for p256k1.

[sergey-shandar]

Note: current building process affects all customers of p256k1. The proposal is to move a part of building.rs which generates bindings to as a separate utility project (Rust script). The script should

• [x] clone the latest secp256k1 C library,
• [x] apply patches,
  • [x] remove static from C source code for some selected functions.
  • [x] add external include to a header file with prefixes.
• [x] generate bindings.

We will commit the patched C source code and generated bindings to Git.

We can split the issue into two tasks:

### Tasks
- [ ] https://github.com/Trust-Machines/p256k1/issues/43
- [ ] https://github.com/Trust-Machines/p256k1/issues/44
- [ ] https://github.com/Trust-Machines/p256k1/issues/46

The first task should fix https://github.com/Trust-Machines/core-eng/issues/188

[sergey-shandar]

Note: we can't prebuild binding.rs because function signatures may depend on a target platform. So, every platform should have its own bindings.rs.

[sergey-shandar]

@xoloki should we create another repo for utility tools to get rid of mono repo? If the build tool is flexible enough, we may share it with https://github.com/rust-bitcoin/rust-secp256k1

[xoloki]

@xoloki should we create another repo for utility tools to get rid of mono repo?

It's fine for now, let's focus on other work. We can do this later once we're not in time crunch.

[sergey-shandar]

@xoloki should we create another repo for utility tools to get rid of mono repo?

It's fine for now, let's focus on other work. We can do this later once we're not in time crunch.

Sure, but feel free to shout ideas into comments.

[sergey-shandar]

See issue #50 and #52 for possible improvements.