blockchain: refactor and export header validation checks

[Crypt-iQ]

This PR allows an external library like neutrino to be able to perform contextual and context-free header validation. This is achieved by introducing two new interfaces HeaderCtx (which replaces *blockNode in the relevant functions) and ChainCtx (which replaces *BlockChain in the relevant functions). A library calling into these functions needs to only satisfy the interface to be able to validate any received headers.

[coveralls]

Pull Request Test Coverage Report for Build 5415613990

• 73 of 130 (56.15%) changed or added relevant lines in 6 files are covered.
• 3 unchanged lines in 1 file lost coverage.
• Overall coverage increased (+0.02%) to 55.235%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
blockchain/chainio.go	0	1	0.0%
blockchain/thresholdstate.go	0	1	0.0%
blockchain/blockindex.go	20	29	68.97%
blockchain/validate.go	43	64	67.19%
blockchain/difficulty.go	6	31	19.35%

Files with Coverage Reduction	New Missed Lines	%
blockchain/difficulty.go	3	48.78%

Totals
Change from base Build 5379381875:	0.02%
Covered Lines:	26688
Relevant Lines:	48317

---

💛 - Coveralls

[guggero]

Looks good! Only blocking comments are about the nil checks vs. interfaces (need to also read up on the specifics, when a normal nil check works and when not).

[Crypt-iQ]

Looks good! Only blocking comments are about the nil checks vs. interfaces (need to also read up on the specifics, when a normal nil check works and when not).

You can nil-check an interface, but it's a bit weird.

See here: https://go.dev/play/p/9NVSUoFo2oI
If foo() returns innerFoo() which returns nil, then the nil is converted to a MyIface with a nil value. This will fail the nil-check.

If foo() just calls return nil, then this will succeed the nil-check. I'll double-check the call-sites you mentioned just to be sure though

[guggero]

Looks good! Only blocking comments are about the nil checks vs. interfaces (need to also read up on the specifics, when a normal nil check works and when not).

You can nil-check an interface, but it's a bit weird.

See here: https://go.dev/play/p/9NVSUoFo2oI If foo() returns innerFoo() which returns nil, then the nil is converted to a MyIface with a nil value. This will fail the nil-check.

If foo() just calls return nil, then this will succeed the nil-check. I'll double-check the call-sites you mentioned just to be sure though

Aaah, I see. Thanks for the example. I guess that makes sense. That's why you had that explicit if x == nil { return nil in there, gotcha.

[Crypt-iQ]

I do wonder if there's not a way to get around nil-checking and to instead use some dummy value to compare against

[Crypt-iQ]

Left some comments, this pattern is kind of worrying since I feel like I've seen it before in our codebases, but can't exactly grep for it

[Roasbeef]

LGTM 🌪️

[guggero]

LGTM 🎉

[Crypt-iQ]

Changed FindPreviousCheckpoint to check and return an error from findPreviousCheckpoint

[Roasbeef]

LGTM 🧱