SSL tests have some bug in cluster configuration

[Lorak-mmk]

When trying to see what tests I can enable after implementing reconnect policies (#402 ) I noticed that SslTests.Integration_Cassandra_ReconnectAfterClusterCrashAndRestart (and presumably other SSL tests) fails for unrelated reason:

[----------] 1 test from SslTests
[ RUN      ] SslTests.Integration_Cassandra_ReconnectAfterClusterCrashAndRestart
1765803052612 [DEBUG] (scylla_cpp_driver::logging:177) message: Log level is set to TRACE
ccm> ccm list
ccm> ccm list
ccm> ccm list
ccm> ccm stop
ccm> ccm status
ccm> ccm create --scylla -n 1:0 -i 127.0.0. -v release:2025.3 -b --ssl=/tmp/ssl cpp-driver_2025-3-0_1-0-ssl
/home/karolbaryla/repos/cpp-rust-driver/tests/src/integration/integration.cpp:192: Failure
Failed
Traceback (most recent call last):
  File "/nix/store/7s0dd9crwvx17pccp2r7w89vdwqcnz2m-scylla_ccm-0.1/bin/..ccm-wrapped-wrapped", line 9, in <module>
    sys.exit(main())
             ^^^^^^
  File "/nix/store/7s0dd9crwvx17pccp2r7w89vdwqcnz2m-scylla_ccm-0.1/lib/python3.11/site-packages/ccmlib/bin/__init__.py", line 75, in main
    cmd.run()
  File "/nix/store/7s0dd9crwvx17pccp2r7w89vdwqcnz2m-scylla_ccm-0.1/lib/python3.11/site-packages/ccmlib/cmds/cluster_cmds.py", line 258, in run
    cluster.enable_ssl(self.options.ssl_path, self.options.require_client_auth)
  File "/nix/store/7s0dd9crwvx17pccp2r7w89vdwqcnz2m-scylla_ccm-0.1/lib/python3.11/site-packages/ccmlib/scylla_cluster.py", line 243, in enable_ssl
    raise FileNotFoundError(f"Required SSL file not found: {file_path}")
FileNotFoundError: Required SSL file not found: /tmp/ssl/ccm_node.pem

[  FAILED  ] SslTests.Integration_Cassandra_ReconnectAfterClusterCrashAndRestart (14525 ms)
[----------] 1 test from SslTests (14525 ms total)

Initially for some reason I thought that a slash is replaced with space for some reason in --ssl flag. But now I see that /tmp/ssl is the path, and cpp-driver_2025-3-0_1-0-ssl is a cluster name, a separate argument.
So i don't really know what is the root cause. Is it a difference between cassandra ccm and our ccm? Is it a bug or something not implemented in our ccm wrapper? Not sure.