Ensure only one query per language is tagged lines-of-code

Some languages have multiple `summary` queries for lines of code,
representing different forms of counting (user written, total, etc).
When Code Scanning sees results from multiple such summary queries in a single run,
it will need to choose one as the primary LoC count to display in the UI.

By ensuring only one query per language has the `lines-of-code` tag,
in future we can teach Code Scanning to look for this particular tag
to identify the primary LoC count.

If a "lines of user code" query is available, use that.
Otherwise use the total "lines of code".

(It is completely fine for multiple queries to be tagged with `summary`.)

cpp/ql/src/Summary/LinesOfCode.ql

@@ -4,7 +4,6 @@
  * @description The total number of lines of C/C++ code across all files, including system headers, libraries, and auto-generated files. This is a useful metric of the size of a database. For all files that were seen during the build, this query counts the lines of code, excluding whitespace or comments.
  * @kind metric
  * @tags summary
- *       lines-of-code
  */
 
 import cpp