Invalid tenure-start block: bad signer signature, tenure_id: 4630b863aedb9ea92b92abf7ddea09fae014a0c1

[pc-quiknode]

Describe the bug
Seeing this error on our mainnet follower nodes:

WARN [1751258312.245595] [stackslib/src/net/download/nakamoto/tenure_downloader.rs:233] [p2p:(20444,20443)] Invalid tenure-start block: bad signer signature, tenure_id: 4630b863aedb9ea92b92abf7ddea09fae014a0c1, block.header.block_id: d9a57d9d61f757c08aa2a0a0dd6206bbb7081369f6b24373839949293d5d9580, state: GetTenureStartBlock(d9a57d9d61f757c08aa2a0a0dd6206bbb7081369f6b24373839949293d5d9580, 1751258311848), error: Public key 0227e324721b149302542657ca3499fda145c1f62e69e236e8bc09ff9c4fb56ac9 not found in the reward set

Environment (please complete the following information):
Version of the appropriate binary / software package:

INFO [1751258543.747198] [testnet/stacks-node/src/main.rs:278] [main] stacks-node 3.1.0.0.12 (release/3.1.0.0.12:ae969ca+, release build, linux [x86_64])
stacks-node 3.1.0.0.12 (release/3.1.0.0.12:ae969ca+, release build, linux [x86_64])

[kantai]

Thanks for reporting this.

This looks like it could be an off-by-one error in the TenureDownloader, which could impact block downloading at reward cycle boundaries -- nodes have other ways of getting block data (push and another path for data pulls), which I think has prevented this off-by-one error from stalling nodes in most cases. However, I think that nodes definitely could be impacted by this.

I'll take a look at getting a repeatable test for this in place, and then open a fix.