Post-fix for MDEV-12253: Remove redundant log writes

The 8 bytes at FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION
will be overwritten at page flush, in buf_page_encrypt_before_write(),
ever since commit 765a4360
(MariaDB 10.1.23, 10.2.6, 10.3.1).

btr_store_big_rec_extern_fields(): Remove useless writes to the
page header (and to the redo log) for ROW_FORMAT=COMPRESSED BLOB pages.

storage/innobase/btr/btr0cur.cc

@@ -7109,33 +7109,6 @@ btr_store_big_rec_extern_fields(
 				/* Initialize the unused "prev page" pointer */
 				mlog_write_ulint(page + FIL_PAGE_PREV,
 						 FIL_NULL, MLOG_4BYTES, &mtr);
-				/* Write a back pointer to the record
-				into the otherwise unused area.  This
-				information could be useful in
-				debugging.  Later, we might want to
-				implement the possibility to relocate
-				BLOB pages.  Then, we would need to be
-				able to adjust the BLOB pointer in the
-				record.  We do not store the heap
-				number of the record, because it can
-				change in page_zip_reorganize() or
-				btr_page_reorganize().  However, also
-				the page number of the record may
-				change when B-tree nodes are split or
-				merged.
-				NOTE: FIL_PAGE_FILE_FLUSH_LSN space is
-				used by R-tree index for a Split Sequence
-				Number */
-				ut_ad(!dict_index_is_spatial(index));
-
-				mlog_write_ulint(page
-						 + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION,
-						 space_id,
-						 MLOG_4BYTES, &mtr);
-				mlog_write_ulint(page
-						 + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION + 4,
-						 rec_page_no,
-						 MLOG_4BYTES, &mtr);
 
 				/* Zero out the unused part of the page. */
 				memset(page + page_zip_get_size(page_zip)