* Release 0.5.7

Raises the tsify-macros floor to 0.5.7 alongside the version bump. The
published 0.5.6 macro predates the Clone derive on JsType, which Ts<T>'s
Clone and its IntoWasmAbi for &Ts<T> both require, so a root-only bump
would let an existing lockfile keep resolving a macro that cannot build
`&Ts<T>` parameters.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* Fix the borrowed-handle leak and correct the release docs

Review of the release candidate turned up a leak in the type whose purpose
is to avoid them: IntoWasmAbi for &Ts<T> cloned the JS handle, but the shim
generated for a borrowed argument never releases it. wasm-bindgen's own
extern types pass the handle by reference for exactly this reason; match
that. The Clone bound falls away with it.

Docs corrected against what the code actually does: only from_wasm_abi
reaches throw_str (into_wasm_abi panics), and the README's namespace enum
and .d.ts examples predated the output they describe. Both are now what a
build emits.

The e2e gate was weaker than it looked: build_all.sh let a failed
wasm-pack build through, compare_output.sh skipped references whose
counterpart was missing, and a generics test asserted the wrong type.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* Drop the changelog entry for a never-released fix, match the sample output

The &Ts<T> leak was introduced and fixed between releases, so listing it
implies 0.5.6 shipped it. "Addresses" instead of "Resolves": #47 keeps an
upstream remainder and #86 an ergonomics one. The .d.ts sample now matches
a build byte for byte.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>