Make MuJoCo Python bindings compatible with free-threading.

Introduce a new header `gil.h` defining `MutexLockIfGilDisabled` to support thread-safety in both standard and free-threaded CPython builds.

Protect critical shared states and registries:
- Guard global Python callback pointers in `callbacks.cc` using a mutex. Move `gil_scoped_acquire` into local blocks around refcount modifications to prevent `longjmp` from bypassing destructors.
- Protect raw pointer maps in `structs_wrappers.cc` with static mutexes.
- Replace TOCTOU race in `mjcb_time` initialization with thread-safe `std::call_once`.
- Add synchronization to lazy indexer array cache initialization in `indexers.cc` and `indexer_xmacro.h`.
- Protect vector mutations in `StructListBase::PopulateUpTo` in `structs.h` with a mutex.
- Revert unnecessary atomic changes to threadpool counters.
- Declare free-threading compatibility by passing `pybind11::mod_gil_not_used()` to all extension modules.

Fixes #3259
Fixes #3256
Fixes #2978

PiperOrigin-RevId: 941101502
Change-Id: Iec4ce58afcbc75d4b0be6a9a21fc8a47854242e3

python/mujoco/experimental/studio/native_viewer.cc

@@ -201,7 +201,7 @@ class Viewer {
   std::vector<std::byte> pixels_;
 };
 
-PYBIND11_MODULE(native_viewer_cc, m) {
+PYBIND11_MODULE(native_viewer_cc, m, pybind11::mod_gil_not_used()) {
   pybind11::module_::import("mujoco._structs");
   pybind11::class_<Viewer>(m, "Viewer")
       .def(pybind11::init<const std::string&, int, int, const std::string&>())

python/mujoco/experimental/studio/parser.cc

@@ -45,7 +45,7 @@ py::object Parse(std::string_view filepath) {
 
 }  // namespace mujoco::python
 
-PYBIND11_MODULE(parser, m) {
+PYBIND11_MODULE(parser, m, pybind11::mod_gil_not_used()) {
   pybind11::module_::import("mujoco._structs");
   m.def("parse", &mujoco::python::Parse,
         pybind11::return_value_policy::take_ownership);

python/mujoco/experimental/studio/renderer.cc

@@ -75,7 +75,7 @@ class Renderer {
 
 }  // namespace mujoco::python
 
-PYBIND11_MODULE(renderer, m) {
+PYBIND11_MODULE(renderer, m, pybind11::mod_gil_not_used()) {
   pybind11::module_::import("mujoco._structs");
   pybind11::class_<mujoco::python::Renderer>(m, "Renderer")
       .def(pybind11::init<const std::string&>())

python/mujoco/experimental/studio/sim.cc

@@ -25,7 +25,7 @@ namespace py = pybind11;
 
 using StepControl = mujoco::platform::StepControl;
 
-PYBIND11_MODULE(sim, m) {
+PYBIND11_MODULE(sim, m, pybind11::mod_gil_not_used()) {
   py::module_::import("mujoco._structs");
   m.doc() = "MuJoCo platform simulation bindings for Link.";
 

python/mujoco/experimental/studio/ux.cc

@@ -51,7 +51,7 @@ struct RenderFlags {
   std::array<uint8_t, mjNRNDFLAG> flags = {0};
 };
 
-PYBIND11_MODULE(ux, m) {
+PYBIND11_MODULE(ux, m, pybind11::mod_gil_not_used()) {
   py::module_::import("mujoco._structs");
   py::class_<RenderFlags>(m, "RenderFlags")
       .def(py::init<>())