tracker-extract: Only index text/plain MIME types, never index source code

Tracker isn't and never will be a sourcecode indexer. Source
code is often huge, and it requires some understanding and
tokenization of the contents in order to effectively create
an index. Tracker indexes text documents by copying the
first 1MB of data from each file into an SQLite database.

src/tracker-extract/15-source-code.rule

-[ExtractorRule]
-ModulePath=libextract-text.so
-MimeTypes=text/x-csrc;text/x-c++src;text/x-chdr;text/x-vala;text/x-java;application/javascript;application/x-php;text/x-python;application/x-perl;application/x-shellscript;text/x-fortran;text/x-pascal;
-FallbackRdfTypes=nfo:SourceCode;nfo:PlainTextDocument;
-Graph=tracker:Documents
-Hash=@hash@

src/tracker-extract/90-text-generic.rule → src/tracker-extract/15-text.rule

 [ExtractorRule]
 ModulePath=libextract-text.so
-MimeTypes=text/*
+MimeTypes=text/plain;text/markdown
 FallbackRdfTypes=nfo:Document;nfo:PlainTextDocument;
 Graph=tracker:Documents
 Hash=@hash@

src/tracker-extract/meson.build

@@ -90,7 +90,7 @@ if get_option('ps')
 endif
 
 if get_option('text')
-  modules += [['extract-text', 'tracker-extract-text.c', ['15-source-code.rule', '90-text-generic.rule'], []]]
+  modules += [['extract-text', 'tracker-extract-text.c', ['15-text.rule'], []]]
 endif
 
 if libtiff.found()