Add some comments to make quorum merkle root calculation more clear+ (#2984)

This should avoid future confusion.

src/evo/cbtx.cpp

@@ -155,6 +155,7 @@ bool CalcCbTxMerkleRootQuorums(const CBlock& block, const CBlockIndex* pindexPre
     static std::map<Consensus::LLMQType, std::vector<const CBlockIndex*>> quorumsCached;
     static std::map<Consensus::LLMQType, std::vector<uint256>> qcHashesCached;
 
+    // The returned quorums are in reversed order, so the most recent one is at index 0
     auto quorums = llmq::quorumBlockProcessor->GetMinedAndActiveCommitmentsUntilBlock(pindexPrev);
     std::map<Consensus::LLMQType, std::vector<uint256>> qcHashes;
     size_t hashCount = 0;
@@ -201,6 +202,9 @@ bool CalcCbTxMerkleRootQuorums(const CBlock& block, const CBlockIndex* pindexPre
             const auto& params = Params().GetConsensus().llmqs.at((Consensus::LLMQType)qc.commitment.llmqType);
             auto& v = qcHashes[params.type];
             if (v.size() == params.signingActiveQuorumCount) {
+                // we pop the last entry, which is actually the oldest quorum as GetMinedAndActiveCommitmentsUntilBlock
+                // returned quorums in reversed order. This pop and later push can only work ONCE, but we rely on the
+                // fact that a block can only contain a single commitment for one LLMQ type
                 v.pop_back();
             }
             v.emplace_back(qcHash);

src/llmq/quorums_blockprocessor.cpp

@@ -405,6 +405,7 @@ bool CQuorumBlockProcessor::GetMinedCommitment(Consensus::LLMQType llmqType, con
     return true;
 }
 
+// The returned quorums are in reversed order, so the most recent one is at index 0
 std::vector<const CBlockIndex*> CQuorumBlockProcessor::GetMinedCommitmentsUntilBlock(Consensus::LLMQType llmqType, const CBlockIndex* pindex, size_t maxCount)
 {
     auto dbIt = evoDb.GetCurTransaction().NewIteratorUniquePtr();
@@ -446,6 +447,7 @@ std::vector<const CBlockIndex*> CQuorumBlockProcessor::GetMinedCommitmentsUntilB
     return ret;
 }
 
+// The returned quorums are in reversed order, so the most recent one is at index 0
 std::map<Consensus::LLMQType, std::vector<const CBlockIndex*>> CQuorumBlockProcessor::GetMinedAndActiveCommitmentsUntilBlock(const CBlockIndex* pindex)
 {
     std::map<Consensus::LLMQType, std::vector<const CBlockIndex*>> ret;