app: recognize REFLEX_COMPILE_PROCESSES and REFLEX_COMPILE_THREADS

Control whether multiprocessing is used and the number of processes or threads
that should be used.

This will allow users to opt-in to the new, potentially hazardous,
multiprocessing mode, which results in much faster compiles, but has already
been reverted 4 times. Lets leave the code in this time, but use the thread
pool executor by default.

Limiting the number of threads or processes to 1 can also aid in debugging
issues that arise during compile time.

reflex/app.py

@@ -860,12 +860,17 @@ class App(Base):
         # Use a forking process pool, if possible.  Much faster, especially for large sites.
         # Fallback to ThreadPoolExecutor as something that will always work.
         executor = None
-        if platform.system() in ("Linux", "Darwin"):
+        if platform.system() in ("Linux", "Darwin") and os.environ.get(
+            "REFLEX_COMPILE_PROCESSES"
+        ):
             executor = concurrent.futures.ProcessPoolExecutor(
-                mp_context=multiprocessing.get_context("fork")
+                max_workers=int(os.environ.get("REFLEX_COMPILE_PROCESSES", 0)) or None,
+                mp_context=multiprocessing.get_context("fork"),
             )
         else:
-            executor = concurrent.futures.ThreadPoolExecutor()
+            executor = concurrent.futures.ThreadPoolExecutor(
+                max_workers=int(os.environ.get("REFLEX_COMPILE_THREADS", 0)) or None,
+            )
 
         with executor:
             result_futures = []