Update rules_python to latest 1.8.3 for tensorflow and xla. Doing so would enable us to use latest feature which includes a venv layout. This latest version also introduce stricter hermeticity which was first introduce in 1.7.0 as seen in logs:

https://rules-python.readthedocs.io/en/latest/changelog.html#v1-7.0 (see sys.path order change and bootstraping)

1. Replace native bazel py_binary/py_library rules with rules_python's py_binary and py_library to selected BUILD files affected with updated to avoid %interpreter% error
2. Export PYTHONPATH build_pip_package.py
3. Add new commit sha and version for rules_python 1.8.3
4. Rules_python 1.8.3 added "scope" params to bool_flag and string_flag which is incompatible with bazel 7.7.0 used by TF, for now we can patch it

PiperOrigin-RevId: 866552189

Author: ecalubaquib

MODULE.bazel

@@ -6,12 +6,12 @@ module(name = "jax")
 
 bazel_dep(name = "abseil-cpp", version = "20250814.1", repo_name = "com_google_absl")
 bazel_dep(name = "bazel_features", version = "1.36.0")
-bazel_dep(name = "bazel_skylib", version = "1.8.1")
+bazel_dep(name = "bazel_skylib", version = "1.8.2")
 bazel_dep(name = "flatbuffers", version = "25.2.10", repo_name = "com_github_google_flatbuffers")
 bazel_dep(name = "platforms", version = "1.0.0")
 bazel_dep(name = "pybind11_bazel", version = "2.13.6")
 bazel_dep(name = "rules_cc", version = "0.2.8")
-bazel_dep(name = "rules_python", version = "1.6.1")
+bazel_dep(name = "rules_python", version = "1.8.3")
 
 # TODO: use a released version when available
 bazel_dep(name = "rules_ml_toolchain")