Offchain Phragmén BREAKING. (#4517)

* Initial skeleton for offchain phragmen

* Basic compact encoding decoding for results

* add compact files

* Bring back Self::ensure_storage_upgraded();

* Make staking use compact stuff.

* First seemingly working version of reduce, full of todos

* Everything phragmen related works again.

* Signing made easier, still issues.

* Signing from offchain compile fine 😎

* make compact work with staked asssignment

* Evaluation basics are in place.

* Move reduce into crate. Document stuff

* move reduce into no_std

* Add files

* Remove other std deps. Runtime compiles

* Seemingly it is al stable; cycle implemented but not integrated.

* Add fuzzing code.

* Cleanup reduce a bit more.

* a metric ton of tests for staking; wip 🔨

* Implement a lot more of the tests.

* wip getting the unsigned stuff to work

* A bit gleanup for unsigned debug

* Clean and finalize compact code.

* Document reduce.

* Still problems with signing

* We officaly duct taped the transaction submission stuff. 🤓

* Deadlock with keys again

* Runtime builds

* Unsigned test works 🙌

* Some cleanups

* Make all the tests compile and stuff

* Minor cleanup

* fix more merge stuff

* Most tests work again.

* a very nasty bug in reduce

* Fix all integrations

* Fix more todos

* Revamp everything and everything

* Remove bogus test

* Some review grumbles.

* Some fixes

* Fix doc test

* loop for submission

* Fix cli, keyring etc.

* some cleanup

* Fix staking tests again

* fix per-things; bring patches from benchmarking

* better score prediction

* Add fuzzer, more patches.

* Some fixes

* More docs

* Remove unused generics

* Remove max-nominator footgun

* Better fuzzer

* Disable it ❌

* Bump.

* Another round of self-review

* Refactor a lot

* More major fixes in perThing

* Add new fuzz file

* Update lock

* fix fuzzing code.

* Fix nominator retain test

* Add slashing check

* Update frame/staking/src/tests.rs

Co-Authored-By: Joshy Orndorff <JoshOrndorff@users.noreply.github.com>

* Some formatting nits

* Review comments.

* Fix cargo file

* Almost all tests work again

* Update frame/staking/src/tests.rs

Co-Authored-By: thiolliere <gui.thiolliere@gmail.com>

* Fix review comments

* More review stuff

* Some nits

* Fix new staking / session / babe relation

* Update primitives/phragmen/src/lib.rs

Co-Authored-By: thiolliere <gui.thiolliere@gmail.com>

* Update primitives/phragmen/src/lib.rs

Co-Authored-By: thiolliere <gui.thiolliere@gmail.com>

* Update primitives/phragmen/compact/src/lib.rs

Co-Authored-By: thiolliere <gui.thiolliere@gmail.com>

* Some doc updates to slashing

* Fix derive

* Remove imports

* Remove unimplemented tests

* nits

* Remove dbg

* Better fuzzing params

* Remove unused pref map

* Deferred Slashing/Offence for offchain Phragmen  (#5151)

* Some boilerplate

* Add test

* One more test

* Review comments

* Fix build

* review comments

* fix more

* fix build

* Some cleanups and self-reviews

* More minor self reviews

* Final nits

* Some merge fixes.

* opt comment

* Fix build

* Fix build again.

* Update frame/staking/fuzz/fuzz_targets/submit_solution.rs

Co-Authored-By: Gavin Wood <gavin@parity.io>

* Update frame/staking/src/slashing.rs

Co-Authored-By: Gavin Wood <gavin@parity.io>

* Update frame/staking/src/offchain_election.rs

Co-Authored-By: Gavin Wood <gavin@parity.io>

* Fix review comments

* fix test

* === 🔑 Revamp without staking key.

* final round of changes.

* Fix cargo-deny

* Update frame/staking/src/lib.rs

Co-Authored-By: Gavin Wood <gavin@parity.io>

Co-authored-by: Joshy Orndorff <JoshOrndorff@users.noreply.github.com>
Co-authored-by: thiolliere <gui.thiolliere@gmail.com>
Co-authored-by: Gavin Wood <gavin@parity.io>

substrate/frame/support/src/traits.rs

@@ -24,7 +24,7 @@ use sp_core::u32_trait::Value as U32;
 use sp_runtime::{
 	RuntimeDebug,
 	ConsensusEngineId, DispatchResult, DispatchError,
-	traits::{MaybeSerializeDeserialize, AtLeast32Bit, Saturating, TrailingZeroInput},
+	traits::{MaybeSerializeDeserialize, AtLeast32Bit, Saturating, TrailingZeroInput, Bounded},
 };
 use crate::dispatch::Parameter;
 use crate::storage::StorageMap;
@@ -87,7 +87,7 @@ impl<
 	Created: Happened<K>,
 	Removed: Happened<K>,
 	K: FullCodec,
-	T: FullCodec
+	T: FullCodec,
 > StoredMap<K, T> for StorageMapShim<S, Created, Removed, K, T> {
 	fn get(k: &K) -> T { S::get(k) }
 	fn is_explicit(k: &K) -> bool { S::contains_key(k) }
@@ -138,6 +138,35 @@ impl<
 	}
 }
 
+/// Something that can estimate at which block the next session rotation will happen. This should
+/// be the same logical unit that dictates `ShouldEndSession` to the session module. No Assumptions
+/// are made about the scheduling of the sessions.
+pub trait EstimateNextSessionRotation<BlockNumber> {
+	/// Return the block number at which the next session rotation is estimated to happen.
+	///
+	/// None should be returned if the estimation fails to come to an answer
+	fn estimate_next_session_rotation(now: BlockNumber) -> Option<BlockNumber>;
+}
+
+impl<BlockNumber: Bounded> EstimateNextSessionRotation<BlockNumber> for () {
+	fn estimate_next_session_rotation(_: BlockNumber) -> Option<BlockNumber> {
+		Default::default()
+	}
+}
+
+/// Something that can estimate at which block the next `new_session` will be triggered. This must
+/// always be implemented by the session module.
+pub trait EstimateNextNewSession<BlockNumber> {
+	/// Return the block number at which the next new session is estimated to happen.
+	fn estimate_next_new_session(now: BlockNumber) -> Option<BlockNumber>;
+}
+
+impl<BlockNumber: Bounded> EstimateNextNewSession<BlockNumber> for () {
+	fn estimate_next_new_session(_: BlockNumber) -> Option<BlockNumber> {
+		Default::default()
+	}
+}
+
 /// Anything that can have a `::len()` method.
 pub trait Len {
 	/// Return the length of data type.