Okay, here's what I understood:

• You opened the channel of 4M, likely in a high fee environment
• the channel opening transaction never made it into the mempool, likely because your node purged it from your mempool (neither your local nor the remote txid can be found in mempool.space, which has a large mempool)
• since channel openings with tx longer than 2 weeks ago, LND purges the opening tx as well
• so you went ahead and force-closed the channel, which never got opened, and now it's borked

All suggestions come with a disclaimer, since it's a bit of an edge case, but many runners had this due to high fee May: some things have worked, some may not. Be a bit cautious on following my remote ideas:

• doublecheck whether you have the opening tx still as unspent in LND: lncli listunspent --unconfirmed_only
• if not, do a SUM comparison between all unspent UTXOs and your wallet: SUM of lncli listunspent vs SUM of lncli walletbalance. It should be pretty equal in sats.
• if those are the same, it's likely those sats never left your wallet. What should have been done: before the end of the 2 weeks, you should have restarted your bitcoin instance with 1GB mempool (300MB is standard), and then RBF your opening transaction. This would be the setting in bitcoin.conf:

# Constrain the mempool to the number of megabytes needed:
#maxmempool=1000

• IF the sats of unspent AND wallet are the same, you could validate those are really yours and no other mempool have mined your transaction by consolidating all UTXOs into one: lncli sendcoins --addr X --sweepall --sat_per_vbyte 10 (or some vByte of your preference), where X is created via lncli newaddress

If this went through, your channel might get discarded in the next two weeks. Otherwise please come back here, if that all doesn't work, guggero's chantools might be needed to sweep committed funds. But as I read it, you have force-closed the channel, so it might not be needed.