add TokenFilter support

[rustmailer]

Hey, could you add TokenFilter support to LinderaTokenizer so it can wrap something like SimpleTokenizer? That way, we could use it like below， making it better for mixed English-Chinese/Korean/Japanese text without changing LinderaTokenStream.

impl Default for TokenizerManager {
    /// Creates an `TokenizerManager` prepopulated with
    /// the default pre-configured tokenizers of `tantivy`.
    fn default() -> TokenizerManager {
        let manager = TokenizerManager::new();
        manager.register("raw", RawTokenizer::default());
        manager.register(
            "default",
            TextAnalyzer::builder(SimpleTokenizer::default())
                .filter(RemoveLongFilter::limit(40))
                .filter(LowerCaser)
                .build(),
        );
        manager.register(
            "en_stem",
            TextAnalyzer::builder(SimpleTokenizer::default())
                .filter(RemoveLongFilter::limit(40))
                .filter(LowerCaser)
                .filter(Stemmer::new(Language::English))
                .build(),
        );
        manager.register("whitespace", WhitespaceTokenizer::default());
        manager
    }
}