Use a predictable order for --lib-search-root

Previously we were using the order in which directory entries are
returned by the filesystem, which is arbitrary, and which can change as
files or directories are added, moved, or removed.

It's more helpful for our users, and easier to document, if we sort the
list of discovered directories.

Signed-off-by: Matt Wozniski <mwozniski@bloomberg.net>

src/pystack/__main__.py

@@ -342,19 +342,13 @@ def process_core(parser: argparse.ArgumentParser, args: argparse.Namespace) -> N
     if args.lib_search_path:
         lib_search_path = args.lib_search_path
     if args.lib_search_root:
-        # We use a dictionary instead of a set to preserve the order of insertion while only
-        # keeping unique path entries in the resulting path
-        so_files = {
-            str(file.parent): None
-            for file in pathlib.Path(args.lib_search_root).glob("**/*.so")
-        }
-        so_files.update(
-            {
-                str(file.parent): None
-                for file in pathlib.Path(args.lib_search_root).glob("**/*.so.*")
-            }
-        )
-        lib_search_path = ":".join(so_files)
+        library_dirs = set()
+        for pattern in {"**/*.so", "**/*.so.*"}:
+            library_dirs.update(
+                str(file.parent)
+                for file in pathlib.Path(args.lib_search_root).glob(pattern)
+            )
+        lib_search_path = ":".join(sorted(library_dirs))
     LOGGER.info("Using library search path: %s", lib_search_path)
 
     corefile_analyzer = CoreFileAnalyzer(corefile, executable, lib_search_path)