Fix non-deterministic mempool reorg test

[pinheadmz]

This is the correct approach to fixing this issue, replacing #518 which still left us with non-deterministic outcomes.

I realized the problem is that the test called fakeClaim() TWICE, meaning the stub resolver actually requests cloudflare's DNSSEC chain twice from the actual legacy DNS. The non-determinism is a result of the stub resolver getting a slightly different result the second time, in particular the RRSIG over the TXT RRs in the final zone. Likely there are many redundant resolvers caching cloudflare.com's TXT RRs and the second fakeClaim() call sometimes ended up getting a slightly newer signature, meaning its inception value exceeded the current chain tip time in the test, and failed with bad-claim-time.

Here's the output of the two fakeClaim() proof data from the original test -- you can see the inception and expiration fields are different -- this is always the case when the test fails. Sometimes the test will succeed if it actually got these two signatures in reverse order (the bug is only thrown if the second proof is newer).

ProofData {
  name: 'cloudflare',
  target: 'cloudflare.com.',
  weak: true,
  commitHash: <Buffer 31 c6 30 af 33 54 f0 3f 20 cc 13 8b 93 84 37 2a 2d 04 b6 d1 06 c2 38 33 f8 d1 83 ec 60 c0 57 73>,
  commitHeight: 1,
  inception: 1609966800,
  expiration: 1610123176,
  fee: 984,
  value: 6800503513487,
  version: 0,
  hash: <Buffer f9 e1 b7 dd 35 34 d5 b6 c6 f8 7f e2 0a d8 89 e5 a5 b2 2d b6>
}
ProofData {
  name: 'cloudflare',
  target: 'cloudflare.com.',
  weak: true,
  commitHash: <Buffer 58 89 27 05 1a 20 a3 c1 2a ff a5 cc f0 d7 f2 ce 18 90 e9 71 e0 e5 66 1f c8 46 7d fe f1 35 5b 44>,
  commitHeight: 120,
  inception: 1609992000,
  expiration: 1610123310,
  fee: 984,
  value: 6800503513487,
  version: 0,
  hash: <Buffer 60 1c 88 fc a7 a0 6e 97 62 3b 5d 45 93 13 e4 b4 a5 15 15 05>
}

The new approach is to only request the DNSSEC chain once at the start of the test, and then modify the commitHash and commitHeight fields once the test chain time is already in the valid range.

[coveralls]

Pull Request Test Coverage Report for Build 478431252

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 1 unchanged line in 1 file lost coverage.
• Overall coverage decreased (-0.005%) to 59.417%

---

Files with Coverage Reduction	New Missed Lines	%
lib/protocol/consensus.js	1	82.79%

Totals
Change from base Build 478034033:	-0.005%
Covered Lines:	19430
Relevant Lines:	30472

---

💛 - Coveralls

[pinheadmz]

[rithvikvibhu]

OS: Manjaro Linux 20.2
Node.js: 14.15.1
hs1qjafwwfv630ldnsveh83lq2zak062r5494zgak3

All tests pass!

[pinheadmz]

hs1qjafwwfv630ldnsveh83lq2zak062r5494zgak3

Sent 01ddc94673ef693f7809222060bbae49d74067899a2f06a2c341cd8dc40b2796 thank you!

[tuxcanfly]

Tested. Can confirm that it cuts down the time by about half 👍