Assuming that the root source of secrecy is only an average password, then the choice of cipher does not matter so much. Especially if the key derivation step is not handled well.
Direct your computing power to:
- Guess a password
- Derive key material using the system parameters
- Use that key to attempt decryption of the ciphertext
- Authenticated encryption provides an oracle for verifying the key guess
This seems like a nice candidate. It uses PBKDF2, which is better than nothing, but it isn't as resource intensive as bcrypt/scrypt/Argon2. It's implemented in python so it shouldn't be too hard to make any necessary modifications.
You can tune the iterations so that it is breakable with your given amount of computing power. As long as the target password isn't too complex, it should be feasible.
How to guess passwords?
There is plenty of software out there already. Some of it will even have nice features like organized word lists, including commonly used passwords and passwords from real breaches.
Depending on how handy you are as a programmer, it may or may not be easy to write your own basic password cracker.