Fix node syncing

[ales-tsurko]

Currently our nodes panic (1 2) when we try to sync them from scratch. The issue is happened because of multiple issues:

• we introduced a runtime upgrade without incrementing the chain spec version (Update testnet #99);
• we patched pallet_democracy and because our nodes couldn't produce blocks without the patch, we used codeSubstitutes in the chainspec.

All those upgrades happened at the same spec version and without spec version update. And it turned out, it's only possible to have a single substitute per spec version, thus using another substitute for the first upgrade provides random substitute usage (either the first or the second one), but not both.

This PR provides:

• switching to WasmExecutor as suggested in 2 as NativeElseWasm may lead to similar issues;
• switching to our fork of polkadot-sdk, where WasmSubstitutes can hold multiple substitutes per the same spec version;
• switching to our fork of frontier, because it depends on polkadot-sdk repo, but we need it to use our fork (it's easier to update Cargo.toml in the fork, than find out which dependencies we need to patch);
• adding bad blocks and fork blocks chainspec extensions;
• updating the chainspec file for testnet with code substitutes;
• removing pallet-democracy-patched as the fix is now in the forked substrate.

Intention behind the fork is caused by the lack of related substrate API, which is neither parametrized nor public. Particularly WasmSubstitute and parts of WasmSubstitutes which holds a map with the spec version as a key and the substitute as a value.