Unescape "*" to "*" in `mw.uri.anchorEncode()`

[xxyzz]

Still don't know how MediaWiki implements anchorEncode, I found this function could unescape "*" in Wiktionary's Lua debug console.

This change fixes the "The specified language Proto-Turkic is unattested, while the given word is not marked with '*' to indicate that it is reconstructed." Lua errors in "Reconstruction" pages

MediaWiki code: https://github.com/wikimedia/mediawiki-extensions-Scribunto/blob/755b549fe66628a2891e9a61a9abade238dd0e9b/includes/Engines/LuaCommon/UriLibrary.php#L29-L33 https://github.com/wikimedia/mediawiki/blob/6592072169f1c25d43723e0956c701855aa4c6ab/includes/parser/CoreParserFunctions.php#L1058-L1062

Lua error: https://kaikki.org/dictionary/All%20languages%20combined/errors/details-The-specified-language-Proto-Turkic-is-yS4Aqcfj.html#LUA-error-in--invoke--links-templates----l_term_t----notself-1---parent---Template-l-self----1---trk-pro---2------usgay-----

More details:

Lua code: https://en.wiktionary.org/wiki/Module:links#L-365--L-375

-- Find embedded links and ensure they link to the correct section.
local function process_embedded_links(text, data, plain)
    -- Process the non-linked text.
    text = data.lang:makeDisplayText(text, data.sc[1], true)  -- "*" is escaped to "*" at here

    -- If the text begins with * and another character, then act as if each link begins with *. However, don't do this if the * is contained within a link at the start. E.g. `|*[[[foo]]](https://en.wiktionary.org/wiki/foo)` would set all_reconstructed to true, while `|[[[*foo]]](https://en.wiktionary.org/wiki/*foo)` would not.
    local all_reconstructed = false
    if not plain then
        -- anchorEncode removes links etc.
        if anchorEncode(text):sub(1, 1) == "*" then   -- anchorEncode convert "*" to "*"
            all_reconstructed = true