This is step (1) in supporting binary in both YAML and non-
YAML files.

This makes it so that instead of immediately converting the
base64-encoded binary into a binary value in python, we just
interpret the binary as a normal string, but annotate it as
such with the `is_binary` flag.

This is needed so that plugins can scan a different value
from the value hashed into baselines.

This implements support for high-entropy secrets in binary
values in yaml files.

We encode the binary value into a hex- or base64-encoded
string (based on the plugin), and run the normal entropy
check. If the string is deemed to be high-entropy, we re
encode the string into a yaml binary (using `yaml.dump`)
and strip the `!!binary`. This yaml binary is considered
the secret, and is put into the baseline as normal.

I had to update a test function so that it uses a custom
hex high-entropy detector, since `HighEntropyStringsPlugin`
is now an abstract class.

To test this you would need to import an unused class
into the module so that it's in `globals()`, and have
the test know what class that is. Seems messy to me,
and not worth what it would be testing.