Not quite. Connection attempts or a majority of packets could be lost that can cause the client-browser to attempt fall-back to HTTP protocol. It is entirely possible in note of this for exit-nodes to entice clients to downgrade to HTTP for malicious purposes but could also happen without elements of malice due to connection/packet-loss between the client, intermediate nodes and the destination server.
If any of the nodes are having issues forcing a new identity may use a different circuit or exit-node that may not have the same issue leading to client-side downgrade of protocol. Oftentimes attempting a reload without new-identity may create a new connection in existing circuit that may work correctly if the fault was intermittent and not malice.
Have reproduced this issue with Tor Browser outside of Whonix. In general if a majority of TLS connections to clear-net sites work correctly any failures are either network faults causing client-side downgrade or poor attempts at protocol-downgrade by attackers. If new identity works to ‘resolve’ the problem would not be too concerned about it.
Sometimes if the archive.org server is slow browsing snapshots from the wayback machine can presumably cause connection timeouts due to the server not responding which might cause the browser to prompt fallback.
But seeing how it is intermittent, is not an automatic downgrade and usually goes away after a reload or new circuit not too concerned about it.