Header refactoring: tsd - cleanup and dependency breaking.

This removes the tsd macros (which are used only for tsd_t in real builds).  We
break up the circular dependencies involving tsd.

We also move all tsd access through getters and setters.  This allows us to
assert that we only touch data when tsd is in a valid state.

We simplify the usages of the x macro trick, removing all the customizability
(get/set, init, cleanup), moving the lifetime logic to tsd_init and tsd_cleanup.
This lets us make initialization order independent of order within tsd_t.

include/jemalloc/internal/tsd_tls.h

@@ -0,0 +1,59 @@
+#ifdef JEMALLOC_INTERNAL_TSD_TLS_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_TLS_H
+
+extern __thread tsd_t tsd_tls;
+extern pthread_key_t tsd_tsd;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	if (pthread_key_create(&tsd_tsd, &tsd_cleanup) != 0) {
+		return true;
+	}
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	/* Do nothing. */
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	return tsd_boot0();
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return false;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	assert(tsd_booted);
+	return &tsd_tls;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	assert(tsd_booted);
+	if (likely(&tsd_tls != val)) {
+		tsd_tls = (*val);
+	}
+	if (pthread_setspecific(tsd_tsd, (void *)(&tsd_tls)) != 0) {
+		malloc_write("<jemalloc>: Error setting tsd.\n");
+		if (opt_abort) {
+			abort();
+		}
+	}
+}