Removed warnings about format strings when running littlefs tests

The difference gcc targets differ with format strings in some odd
(but not against the rules) ways. Most notable, the uint32_t in
i386/x86_64 gcc uses %d, whereas on cortex-m gcc uses %ld.

This makes dealing with warnings on code that goes between the two
rather annoying.

.travis.yml

@@ -11,23 +11,23 @@ script:
           -n 'tests*'
 
       # Run littlefs functional tests
-    - CFLAGS="-Wno-error=format" make -Clittlefs test QUIET=1
+    - CFLAGS="-Wno-format" make -Clittlefs test QUIET=1
 
       # Run littlefs functional tests with different configurations
       # Note: r/w size of 64 is default in mbed
-    - CFLAGS="-Wno-error=format -DLFS_READ_SIZE=64 -DLFS_PROG_SIZE=64"
+    - CFLAGS="-Wno-format -DLFS_READ_SIZE=64 -DLFS_PROG_SIZE=64"
             make -Clittlefs test QUIET=1
-    - CFLAGS="-Wno-error=format -DLFS_READ_SIZE=1 -DLFS_PROG_SIZE=1"
+    - CFLAGS="-Wno-format -DLFS_READ_SIZE=1 -DLFS_PROG_SIZE=1"
             make -Clittlefs test QUIET=1
-    - CFLAGS="-Wno-error=format -DLFS_READ_SIZE=512 -DLFS_PROG_SIZE=512"
+    - CFLAGS="-Wno-format -DLFS_READ_SIZE=512 -DLFS_PROG_SIZE=512"
             make -Clittlefs test QUIET=1
-    - CFLAGS="-Wno-error=format -DLFS_BLOCK_COUNT=1023"
+    - CFLAGS="-Wno-format -DLFS_BLOCK_COUNT=1023"
             make -Clittlefs test QUIET=1
-    - CFLAGS="-Wno-error=format -DLFS_LOOKAHEAD=2048"
+    - CFLAGS="-Wno-format -DLFS_LOOKAHEAD=2048"
             make -Clittlefs test QUIET=1
 
       # Self-host with littlefs-fuse for fuzz test
-    - make -C littlefs-fuse
+    - CFLAGS="-Wno-format" make -C littlefs-fuse
 
     - littlefs-fuse/lfs --format /dev/loop0
     - littlefs-fuse/lfs /dev/loop0 mount
@@ -37,7 +37,7 @@ script:
     - cp -r $(git ls-tree --name-only HEAD littlefs/) mount/littlefs
     - cd mount/littlefs
     - ls
-    - CFLAGS="-Wno-error=format" make -B test_dirs QUIET=1
+    - CFLAGS="-Wno-format" make -B test_dirs QUIET=1
 
 install:
       # Get arm-none-eabi-gcc