Improve SignedExtension matching logic and remove SkipCheckIfFeeless bits

[jsdw]

Closes #1280.

• SkipCheckIfFeeless is now transparent to the outside world in the latest polkadot-sdk, so remove logic specific to that.
• But, remove SignedExtension::IDENTIFIER and replace with SignedExtension::matches(..) function which is able to do much more custom/intelligent matching to work out which of our signed extensions to use at which point (in part, just in case we have more complex things to do later, and in part just to provide the flexibility now for anybody that wants it).
• Tidy some bounds (Debug/Clone only for the signed exts the user might handle ie the ones that are Decode targets) and no need for EncodeAsType/DecodeAsType on most of the exts now.
• Just return ExtrinsicParamsError rather than type Error: Into<ExtrinsicParamsError>.

[niklasad1]

why removed?

[jsdw]

it would never have worked anyway afaik, because T: Config doesn't impl Debug. Maybe I was a bit overzealous in removing a bunch of derives for these things though; I did initially replace them all with working ones and then backed out of it!

[niklasad1]

I like it, it simplifies the signed extension impls.

[tadeohepperle]

I don't really understand the reasons why SkipCheckIfFeeless was introduced and why we don't need it as a singed extension in subxt anymore.

[niklasad1]

I don't really understand the reasons why SkipCheckIfFeeless was introduced and why we don't need it as a singed extension in subxt anymore.

AFAIU, SkipCheckIfFeeless was introduced as "wrapper signed extension" that wrapped something like "ChargeTxPayment", initially the name of the signed extension was "SkipCheckIfFeeless" but that was reverted to propagate the name of the signed extension that was wrapped.

Such SkipCheckIfFeeLess, its name will be "ChargeTxPayment" and not "SkipCheckIfFeeLess"

[jsdw]

I don't really understand the reasons why SkipCheckIfFeeless was introduced and why we don't need it as a singed extension in subxt anymore.

Yup, a couple of Substrate PRs made the name return the innet signed ext name, and the typeinfo match the inner typeinfo, so SkipCheckIfFeeless is now invisible to us and can be ignored again :)