A transaction is an atomic operation, it completes entirely or not at all. A single-statement delete is a single transaction, no matter how many rows are deleted and as such must succeed or fail as a single operation. If the delete ran out of log space and failed, the entire delete would have to have rolled back.
A manual checkpoint was probably not required, wouldn't have allowed the log to be reused with an active transaction in it. What most likely happened is that the log autogrow didn't happen fast enough and the delete failed and rolled back. Next automatic checkpoint the log was cleared and when you re-ran the delete there was enough space because the log had grown.
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
We walk in the dark places no others will enter
We stand on the bridge and no one may pass