Verifier.__init__ raises unhandled IndexError when the trusted root has no transparency logs

[lsd-cat]

If a trusted_root with no tlog entries is provided, the verfier crashes with a standard python error.

See:

sigstore-python/sigstore/verify/verifier.py

Lines 89 to 91 in 2c74d8f

	# this is an ugly hack needed for verifying "detached" materials
	# In reality we should be choosing the rekor instance based on the logid
	url = trusted_root._inner.tlogs[0].base_url

Since the verifier documents its failures as VerificationError, a caller catching only VerificationError gets an uncaught crash. It's not security relevant both because the crash is blocking, and because it assume a likely malformed trusted_root (or is it supported to do a private sigstore deploy with no logs?).

The other verifiers (js/rs/go) seem to throw different failed verification errors for the same test case.

[woodruffw]

This is intentional -- we expect a Sigstore instance to have at least one tlog. I guess we could turn that into a more structured error, though.