Update chen for improved performance with header analysis

[prabhu]

-l h must be a bit faster now. This is achieved by:

• Not searching for includePaths in header mode. This would result in header files being represented as relative files in usage slices.

{
      "code" : "#include<sys/socket.h>",
      "fullName" : "sys/socket.h",
      "signature" : "sys/socket.h",
      "fileName" : "",
      "lineNumber" : null,
      "columnNumber" : null,
      "usages" : [
      ]
    },

instead of

{
      "code" : "",
      "fullName" : "/usr/include/asm-generic/socket.h:12:12:SO_REUSEADDR:0",
      "signature" : "",
      "fileName" : "/usr/include/asm-generic/socket.h",
      "lineNumber" : 12,
      "columnNumber" : null,
      "usages" : [

If you prefer the depth and detail shown in (2), use -l c.

• Ignoring several nodes in the AST, such as control flow, expressions, and several operators.

[prabhu]

Creating a usage slice using -l h is now finishing in GitHub actions in around an hour. With 32GB heap memory, it is finishing in under 15 minutes.