In order to use the db for system monitoring we need a way of mapping the locator the the uuids the appointments/jobs are identified in the db. First steps towards #15
- Adds a PRG based on a hardcoded seed to make the tests reproducible (get_random_value_hex)
- Updates all the tests replacing urandom for get_random_value_hex
- Properly places misplaced bitcoin_cli in bitcoin_sim_tests
- Typos
bitcoin_cli as a global variable in the main __init__.py was creating issues related to http.client.CannotSendRequest: Request-sent and connection re-usage. Define a new connection per request.
The decryption for the `EncryptedBlob` using AES-GCM-128 (the only cipher available atm) raises an `InvalidTag` exception. This was not properly captured by the watcher making it crash. This behavior was already discovered during the `EncryptedBlob` unit testing and left to be fixed in the `Watcher` unit testing.
However, making the EncryptedBlob raise such an exception may not be a good practice, since other ciphers may run into different exceptions. Therefore, the `EncryptedBlob` has been modified to return None upon facing a decryption issue, the `BlockProcessor` will detect that and return a None justice_txm and justice_txid. Upon receiving a None `justice_txid` the `Watcher` will delete the appointment without notifiying the `Responder`.
The modules that need to run either bitcoind or the api do now reference to the fixture in the first test that needs it. Since the fixtures are definexd session-wise the rest of the modules will have access to them from that point on.
test_get_all_appointments_responder was failing unconsistently. It was due to how the api fixture was set up.
Fix:
- Sets up bitcoind and api fixtures to run session-wise (so they are not re-initialized for every test
- Updates tests accordingly (e.g. reduces the number of MULTIPLE_REQUESTS so it does not cap)
- Keeps track of all sent appointments so the test_request functionds don't need to create additional ones
