Move sr-arithmetic to a new crate and add in a fuzzer (#3799)

* Split up sr_arithmetic.rs

* Add some basic fuzzing

* Add more tests

* Add printing to fuzzing

* Clean things up

* Remove arbitrary

* Remove comments

* More cleaning, fix small error that was causing a panic

* Add rational128

* Remove old random tests

* introduce panic

* fuzzing should panic properly

* Bit of cleanup

* Add a test uncovered via fuzzing that fails!

* Few small changes

* Move sr-arithmetic to its own crate

* Fix fuzzing

* Got rid of fuzzer Cargo.lock

* Added no_std

* re-export assert_eq_error_rate

* bump impl and spec version

* re add convert into

* Add an ignore to the test

* Enabled benchmarking

* Reindent

* Clean up biguint fuzzer

* Clean up biguint more

* shuffle sr-primitives/traits about

* Remove unused dependencies

* Apply clippy suggestions

* upgrade primitive-types versions

* Run tests against num-bigint

* Get rid of allocation in assert_biguints_eq

* Add an optimisation to multiply_by_rational

* rename parts_per_x -> per_things

* Change fuzzer cargo.toml

* Remove allocation from BigUint PartialEq impl

* Remove accidental indentation

* Renmove Lazy and Convert traits

* Copy assert_eq_error_rate macro back to sr-primitives

* Add documentation to fuzzers

* fix sr-primitives assert_eq_error_rate

* add cfg(test)

* Update core/sr-arithmetic/src/traits.rs

Co-Authored-By: Kian Paimani <5588131+kianenigma@users.noreply.github.com>

* Update core/sr-arithmetic/src/traits.rs

Co-Authored-By: Kian Paimani <5588131+kianenigma@users.noreply.github.com>

* Update core/sr-arithmetic/fuzzer/src/biguint.rs

Co-Authored-By: Kian Paimani <5588131+kianenigma@users.noreply.github.com>

* Allow rounding up in rational128

* Make changes to biguint.rs

* Update core/sr-arithmetic/src/traits.rs

Co-Authored-By: Kian Paimani <5588131+kianenigma@users.noreply.github.com>

* Final touches

* Convert to num_bigint::BigUint to compare

* remove unused mut

* more small changes

* shuffle sr-primitives trait imports

* more code review

* move assert_eq_error_rate to lib.rs

* Update core/sr-arithmetic/fuzzer/src/biguint.rs

Co-Authored-By: Bastian Köcher <bkchr@users.noreply.github.com>

* Get rid of S

* Simplify rational128 honggfuzz link

* Insignificantly change rational128 fuzzing code

* Slightly tidy up some of the arithmetic logic

* Get rid of sr_arithmetic again(?) and fix sr-primitives/weights

* Apply updates to sr_arithmetic.rs to crate

substrate/core/sr-primitives/src/lib.rs

@@ -51,7 +51,6 @@ pub mod testing;
 pub mod curve;
 pub mod generic;
 pub mod offchain;
-pub mod sr_arithmetic;
 pub mod traits;
 pub mod transaction_validity;
 pub mod weights;
@@ -64,14 +63,14 @@ pub use primitives::{TypeId, crypto::{key_types, KeyTypeId, CryptoType}};
 pub use app_crypto::RuntimeAppPublic;
 
 /// Re-export top-level arithmetic stuff.
-pub use sr_arithmetic::{
+pub use arithmetic::{
 	Perquintill, Perbill, Permill, Percent,
 	Rational128, Fixed64
 };
 /// Re-export 128 bit helpers.
-pub use sr_arithmetic::helpers_128bit;
-/// Re-export big_uint stiff.
-pub use sr_arithmetic::biguint;
+pub use arithmetic::helpers_128bit;
+/// Re-export big_uint stuff.
+pub use arithmetic::biguint;
 
 /// An abstraction over justification for a block's validity under a consensus algorithm.
 ///