A lot of noise is often included in jextract artifacts, because libraries depend on system headers, which are then transitively included in full.

To reduce the size of the jextract artifacts, after some discussing: https://mail.openjdk.java.net/pipermail/panama-dev/2019-March/004828.html & https://mail.openjdk.java.net/pipermail/panama-dev/2019-March/004932.html

We have come up with the following plan.

1.) Start from a root set of header files H, which contains all headers, filtered by some patterns passed to --include-headers & --exclude-headers.

2.) Include anything found in the headers in H.

3.) Also include elements (but not entire headers) that are required by something appearing in H (based on dependency analysis).