multipart: expose parsed Content-Disposition params on file/field info

[Martin-Luther]

Adds the already-parsed Content-Disposition parameters to the info object that
is passed to the file and field events, as info.dispositionParams.

Motivation

A multipart part's Content-Disposition header can carry parameters beyond name
and filename. busboy already parses the complete parameter list via parseParams,
but only surfaces name and filename — so a consumer that needs any other
parameter has to re-parse the raw header itself.

That duplication is the problem: a second parser has to re-implement RFC 2231
continuations and charset-tagged values (name*=utf-8''...) to stay consistent with
busboy, and any divergence between the two silently misattributes parts. Exposing the
value busboy has already computed removes the need for a second parser entirely.

What changes

• lib/types/multipart.js: the parsed disp.params is retained per part and included
  in the info object emitted with the file and field events, as
  dispositionParams. It is null for a part with no Content-Disposition header.

That is the whole change — 10 lines. No new options, no signature changes, nothing
removed or renamed.

Compatibility

Strictly additive: one new property on an object that is already passed to the
listener. info.name / info.filename / info.encoding / info.mimeType and every
existing event signature are untouched, so existing consumers are unaffected whether
or not they read the new property.

Tests

• test/test-types-multipart-disposition-params.js (new) — covers the exposed
  parameters for both file and field parts, including a part with no
  Content-Disposition header.
• test/test-types-multipart.js — extended to assert the new property alongside the
  existing expectations.
• test/test-types-multipart-charsets.js, test/test-types-multipart-stream-pause.js
  — updated where they assert the full info shape.

Related

#183 asks for the same underlying capability (making additional multipart header
information available). This PR is the narrower form of that request: rather than
surfacing arbitrary headers, it exposes only what parseParams already produces for
Content-Disposition, which keeps the surface small and avoids any new parsing.