t2u tag parsing scans too much

Consider the tag below.

The lines object ..tagger are parts of the git object and mustn't be scanned for metadata. But my reading of the code (that I wrote) in dgit-repos-server is that it does scan them. This may not matter because they won't start with [ let alone [dgit.

But, there is another question: should the title line (in this case dgit-test-dummy release 1.40 for experimental) be scanned?

I propose the following answers:

1. We should not scan the title line, because sometimes people put [subject line tag] in things like email subjects.
2. That means we should insist on the 2nd line being blank (I think git doesn't handle a markdown-style multi-line paragraph here) and regard the tag as not-for-t2u otherwise.
3. We should not scan the git data fields in the git object header.

Please let me know if you agree, and make whatever consequential changes to dgit.git?

Thanks.

object b8e6d37e61b2eebc030752f0f4962960d87c1af4
type commit
tag debian/1.40
tagger Ian Jackson <ijackson@chiark.greenend.org.uk> 1725709192 +0100

dgit-test-dummy release 1.40 for experimental

[dgit distro=debian split]
[dgit please-upload source=dgit-test-dummy version=1.40]
-----BEGIN PGP SIGNATURE-----

iQFUBAABCAA+FiEEVZrkbC1rbTJl58uh4+M5I0i1DTkFAmbcO4ggHGlqYWNrc29u
QGNoaWFyay5ncmVlbmVuZC5vcmcudWsACgkQ4+M5I0i1DTmD4gf/Q8xD7Eic90CV
A8WQV9boU0j0oUEPioIiMqsKO3fHf66xPMukEFtpYcX5KOvPrpRGDYseP1JiPUNM
/mYWfwLXECSkypjidFRvKEMFivvvDFZfJnyqQ6MTIvTsIPYAy/sZnUd/1+v5wcMb
4kGYMBXJJZ749N5voi+wHfZfEHgOmvfOnO7pwk60xgm07lUGHdQCxN7ndIqOHmxg
FJZHJu0ajp4qfx7kqIdwPWxW2gmN4iqzLsTNcOuQAiIKBn2LtetMRfxBHHjoYv2j
1OzZ0OcDXAp/2OkxT3iYmFjTrobOJTlmg3WRGCwqzA/aYz8T8kelqgVXS8uR4tcF
nmuEyTw7YA==
=rZZP
-----END PGP SIGNATURE-----