use folly::hardware_concurrency instead of std::thread::hardware_concurrency

Summary:
This change replaces calls of std::thread::hardware_concurrency with
folly::hardware_concurrency when the caller wants the number of
processors available to a job.

This should be largely a no-change since most jobs use platform010
with glibc 2.34 where these functions behave identically.  However,
jobs on Aarch64 use glibc 2.40.  There, the folly method continues to
return the number of processors available to a job but the std version
returns the number of processors known to the operating system, a
different number when the job is stacked.

Most of fbcode expects the folly behavior and this codemod only
affects those callers.  By making the switch to the folly method, code
will behave the same when code is run on Aarch64 or when the glibc is
upgraded on x86.

Reviewed By: yfeldblum

Differential Revision: D87904531

fbshipit-source-id: c8b29b8ea20d6036c8098d847569edcc1b95254d

Author: lexprfuncall

third-party/wangle/src/wangle/bootstrap/ServerBootstrap.h

@@ -18,6 +18,7 @@
 
 #include <folly/io/async/AsyncTransport.h>
 #include <folly/synchronization/Baton.h>
+#include <folly/system/HardwareConcurrency.h>
 #include <wangle/bootstrap/ServerBootstrap-inl.h>
 #include <wangle/channel/Pipeline.h>
 #include <iostream>
@@ -136,7 +137,7 @@ class ServerBootstrap {
           1, std::make_shared<folly::NamedThreadFactory>("Acceptor Thread"));
     }
     if (!io_group) {
-      auto threads = std::thread::hardware_concurrency();
+      auto threads = folly::hardware_concurrency();
       if (threads <= 0) {
         // Reasonable mid-point for concurrency when actual value unknown
         threads = 8;