and the Condorcet method can completely fail to select a winner.

That one’s not a flaw. All elections can suffer from ties. Pure Condorcet just makes it obvious when there’s a tie (and this is very rare). There are a bunch of Condorcet completion methods for resolving the tie.

Until ranked choice for president

That wouldn’t change anything. RCV still produces a polarized two-party system.

Ranked Choice Voting doesn’t make third parties viable, either. It uses the same counting method as our current system (tally up people’s first-choice preferences) and therefore suffers from all the same problems, like vote-splitting, spoiler effect, and center-squeeze effect. You can’t fix the problems of FPTP by adding more rounds of FPTP. You need to allow voters to express opinions about all of the candidates and then actually count all of those opinions.

If you want third parties to be viable, you want real reforms like STAR Voting, Condorcet RCV, or Approval Voting.

You probably mean https://www.rcv123.org/? But that uses Hare’s method, which is pretty flawed and won’t always pick the candidate preferred by the voters.

Better to use https://star.vote/ (score ballots) or https://civs1.civs.us/civs_create.html (ranked ballots).