Add new config options for fts_solr

[arnisoph]

"From", "To", "Cc", "Bcc", "Subject"

The default for sending header fields to Solr is to send a limited number of headers ("From", "To", "Cc", "Bcc", "Subject") as "field" attributes and all other headers with name and content in the "hdr" field.

<field name="From">[email protected]</field>
<field name="To">[email protected]</field>
..
<field name="hdr">
From: [email protected]
To: [email protected]
...
X-Origin-Reference: foo1
X-Origin: foo2
X-Origin-Recipient: foo3
X-Priority: 3
...
X-Foo: content of x-foo
</field>

The new config switches to the following:
limit_mime_hdr: limit fields listed in "hdr" field to the same fields as in the standard fields header (From, To, Cc, Bcc, Subject):

<field name="From">[email protected]</field>
<field name="To">[email protected]</field>
..
<field name="hdr">
From: [email protected]
To: [email protected]
</field>
skip_mime_hdr_fieldname: Do not put the name of the field in the "hdr" field
<field name="From">[email protected]</field>
<field name="To">[email protected]</field>
..
<field name="hdr">
[email protected]
[email protected]
...
foo1
foo2
foo3
3
...
content of x-foo
</field>

if you set both of the config values the result will be something like this:

<field name="From">[email protected]</field>
<field name="To">[email protected]</field>
..
<field name="hdr">
[email protected]
[email protected]
...
</field>

To use this config values add them to the fts_solr parameter list:

plugin {
  # Search
  fts = solr
  fts_solr = break-imap-search url=http://solr.example.com:8080/solr/collection1/ limit_mime_hdr skip_mime_hdr_fieldname
}

[sirainen]

These changes make using fts-solr even less IMAP-compliant though.. For example I think it now breaks trying to search Message-IDs? That's at least somewhat commonly used by IMAP clients (e.g. iOS is using it excessively nowadays for some weird reason).

[arnisoph]

Timo, thanks for your feedback.

I've discussed with my colleague who created this patch and we came to the conclusion that this patch isn't necessary and the issue behind it needs a solution on our side (instead on Dovecot side).

Closing this PR as described above..