Always generate fast-mode data for NFD and NFKD tries (#7222)

The intent is to allow the normalizer to assume that these tries are
fast in the baked data mode.

This changeset removes the ability to generate serde-loadable data with
small-mode tries for NFD and NFKD. Per previous discussion, we can add
the ability to generate small tries for these back if there are users of
ICU4X who really need that mode for these tries.

Fixes #6836

Author: hsivonen

provider/data/normalizer/data/normalizer_nfd_data_v1.rs.data

@@ -1,7 +1,7 @@
 normalizer/nfc/v1, <singleton>, 5332B, 5310B, dd048e68b718a993
-normalizer/nfd/data/v1, <singleton>, 28208B, 28144B, a37d3c274c030323
+normalizer/nfd/data/v1, <singleton>, 34912B, 34848B, 933447e5c056e49c
 normalizer/nfd/supplement/v1, <singleton>, 4486B, 4403B, a587960aa6faa95c
 normalizer/nfd/tables/v1, <singleton>, 2177B, 2133B, 8818c6d23b22796a
-normalizer/nfkd/data/v1, <singleton>, 43400B, 43336B, 855553ea574c0178
+normalizer/nfkd/data/v1, <singleton>, 51488B, 51424B, a5a4848d32dd13f8
 normalizer/nfkd/tables/v1, <singleton>, 5865B, 5821B, 7f50ce23df00365d
 normalizer/uts46/data/v1, <singleton>, 56552B, 56493B, 2f3422c2c9cc708b

provider/data/normalizer/data/normalizer_nfkd_data_v1.rs.data

@@ -142,7 +142,8 @@ struct Cli {
     #[arg(
         help = "Whether to optimize CodePointTrie data structures for size (\"small\") or speed (\"fast\").\n\
                   Using \"fast\" mode increases performance of CJK text processing and segmentation. For more\n\
-                  information, see the TrieType enum."
+                  information, see the TrieType enum. The tries for the core (UAX #15 but not UAX #46)\n\
+                  normalization forms use the fast trie type regardless of this setting."
     )]
     #[cfg(feature = "provider")]
     trie_type: TrieType,