main:cpp: Fix thread race

There is some a global variable read_time which is updated by
all threads without proper synchronization.

    WARNING: ThreadSanitizer: data race (pid=15896)
      Read of size 4 at 0x00000066d53c by thread T7:
        #0 t_PairAlign(void*) /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:118 (bsmap+0x00000042c4f2)

      Previous write of size 4 at 0x00000066d53c by thread T6:
        [failed to restore the stack]

      Location is global 'read_time' of size 4 at 0x00000066d53c (bsmap+0x00000066d53c)

      Thread T7 (tid=15960, running) created by main thread at:
        #0 pthread_create /scratch/local/bee-root/gcc/gcc-5.3.0-0/source/libsanitizer/tsan/tsan_interceptors.cc:895 (libtsan.so.0+0x000000026c44)
        #1 Do_PairAlign() /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:128 (bsmap+0x00000042c613)
        #2 RunProcess() /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:499 (bsmap+0x0000004306b0)
        #3 main /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:611 (bsmap+0x0000004320ae)

      Thread T6 (tid=15959, finished) created by main thread at:
        #0 pthread_create /scratch/local/bee-root/gcc/gcc-5.3.0-0/source/libsanitizer/tsan/tsan_interceptors.cc:895 (libtsan.so.0+0x000000026c44)
        #1 Do_PairAlign() /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:128 (bsmap+0x00000042c613)
        #2 RunProcess() /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:499 (bsmap+0x0000004306b0)
        #3 main /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:611 (bsmap+0x0000004320ae)

    SUMMARY: ThreadSanitizer: data race /scratch/cluster/buczek/bsmap/bsmap.git/main.cpp:118 t_PairAlign(void*)
    ==================

This is not a real issue, because the value is not currently
used. But because we are holding the mutex_fout mutex at this point
in code anyway, swap the order and work in the global first, then
release the mutex. This puts read_time under the protection of
mutex_fout.

main.cpp

@@ -77,8 +77,8 @@ void *t_SingleAlign(void *tid) {
 	}
 	pthread_mutex_lock(&mutex_fout);
 	n_aligned+=a.n_aligned; n_unique+=a.n_unique; n_multiple+=a.n_multiple;
-	pthread_mutex_unlock(&mutex_fout);
 	read_time+=Cal_AllTime()-ref_time;
+	pthread_mutex_unlock(&mutex_fout);
 	return NULL;
 };
 
@@ -115,8 +115,8 @@ void *t_PairAlign(void *tid) {
 	n_aligned_pairs+=a.n_aligned_pairs; n_unique_pairs+=a.n_unique_pairs; n_multiple_pairs+=a.n_multiple_pairs;
 	n_aligned_a+=a.n_aligned_a; n_unique_a+=a.n_unique_a; n_multiple_a+=a.n_multiple_a;
 	n_aligned_b+=a.n_aligned_b; n_unique_b+=a.n_unique_b; n_multiple_b+=a.n_multiple_b;
-	pthread_mutex_unlock(&mutex_fout);
 	read_time+=Cal_AllTime()-ref_time;
+	pthread_mutex_unlock(&mutex_fout);
 	return NULL;		
 };