This commit adds the `_batch_create` and `_batch_destroy` APIs.

- `_batch_scratch_size` allocation formula is taken from bench ecmult:
https://github.com/bitcoin-core/secp256k1/blob/694ce8fb2d1fd8a3d641d7c33705691d41a2a860/src/bench_ecmult.c#L312
- `aux_rand16` param in `_batch_create` is used for synthetic randomness,
while generating randomizers: sipa/bips#204

This commit adds the batch APIs:

1. batch_add_schnorrsig
    Adds a Schnorr signature to the batch

2. batch_add_xonlypub_tweak_check
	Adds a tweaked x-only pubkey check to the batch

3. batch_usable
	Checks if a batch can be used by _batch_add_* APIs.

**Side Note:**
Exposing batch_add_schnorrsig in the `secp256k1_schnorrsig.h` header
file (with batch module header guards) will force the user to define
ENABLE_MODULE_BATCH during their code compilation. Hence, it is in a
standalone `secp256k1_schnorrsig_batch` header file. A similar argument
could be made for batch_add_xonlypub_tweak_check.

This commit adds an example C program using the batch API.

GNU Autotools will compile this example only if both batch and
schnorrsig modules are enabled.

…ctor

This commit adds the following tests:
	1. Cirrus scripts
	2. Batch API tests (ordered)
	3. Tagged SHA256 test
	4. BIP340 test vectors: https://github.com/bitcoin/bips/blob/master/bip-0340/test-vectors.csv
	5. Large random test for `strauss_batch` refactor

This commit adds the following tests:
	1. Random bitflip test for randomizer generating function
	2. Random bitflip in Schnorr Signature
	3. NULL arg tests

This commit adds benchmarks for Schnorr signature batch verification,Tweaked
pubkey check batch verification, and Tweaked pubkey check (single verification).

For batch verify benchmark, the number of sigs (or checks) in the batch varies
from 1 to SECP256K1_BENCH_ITERS with a 20% increment.

This commit generates graphs that visualize the batch verify speed up over
single verification (y-axis) wrt the number of signatures (or tweak checks)
in the batch. The input data points are taken from the batch verify benchmark.

GNU plot was used to generate these graphs. The instructions to reproduce these
graphs (on your local machine) are given in doc/speedup-batch.md

The value of `STRAUSS_MAX_TERMS_PER_BATCH` was calculated (approx) from
the generated graphs. Relevant discussion: #2 (comment)