Grow dual thread stack after pushing a thread on the next queue

The previous code was assuming it was fine to push_next without
growing, which used to be the case with the previous implementation
because we always have poped the current thread that we try to push.

However now that we use a ring-buffer, m_next_begin == m_next_end can
either mean full, or empty. We solve this by assuming it means empty
and never allowing the buffer to become full, which means we need
to grow after pushing to next if we get full.

Fixes #4859

src/regex_impl.hh

@@ -504,7 +504,6 @@ private:
                 if (start_desc and m_threads.next_is_empty())
                     to_next_start(pos, config, *start_desc);
                 m_threads.push_next({first_inst, -1});
-                m_threads.grow_ifn(false);
             }
             m_threads.swap_next();
         }
@@ -632,7 +631,7 @@ private:
         void push_current(Thread thread) { m_data[decrement(m_current)] = thread; grow_ifn(true); }
         Thread pop_current() { return m_data[post_increment(m_current)]; }
 
-        void push_next(Thread thread) { m_data[post_increment(m_next_end)] = thread; }
+        void push_next(Thread thread) { m_data[post_increment(m_next_end)] = thread; grow_ifn(false); }
         Thread pop_next() { return m_data[decrement(m_next_end)]; }
 
         void swap_next()

test/regression/4859-regex-invalid-behaviour/cmd

@@ -0,0 +1 @@
+ge<a-/><ret>d

test/regression/4859-regex-invalid-behaviour/in

@@ -0,0 +1,3 @@
+             ddddd
+             aaaaa
+             bbbbb

test/regression/4859-regex-invalid-behaviour/out

@@ -0,0 +1 @@
+            bbbbb

test/regression/4859-regex-invalid-behaviour/rc

@@ -0,0 +1 @@
+reg slash '(?S)((^ {13}d.*\n)(^ {13}.*\n)*?)?^.'