Use Biatec Router's combined-split quotes (routesCount: 1) instead of legacy alternatives

biatec.ts requested routesCount: 3 for both getQuote and execute, which forces
BiatecRouter's legacy "N independent full-amount alternatives" mode and disables
its v2 multi-leg splitting (BuildCombinedRoute) entirely - the router can never
split a swap across multiple paths the way Haystack/Deflex routinely does.

Live comparison (1000 VOTE -> EURS): switching to routesCount: 1 (splitting
enabled) took Biatec production from a 12.8% gap vs Deflex down to 6.0%, and
stage from 18.8% down to 8.5% - roughly halving the shortfall for this pair.

With routesCount <= 1, response.routes holds the LEGS of one combined swap, not
independent alternatives: total output is the sum across legs, and the full
transaction group is the concatenation of every leg's txsToSign (all legs share
one Algorand atomic group server-side - signing only routes[0]'s slice would
submit an incomplete group). Added combineRouteResponse() to fold that into the
same { route: { route, txsToSign } } shape the rest of this file and
buildBiatecRouteInfo already expect, so downstream code needs no other changes.

Known limitation: the hop/pool breakdown display still only reflects the first
leg when the router splits - a display gap, not a correctness one (quoted
amount and submitted transactions are both correct across all legs). A follow-up
could teach SwapRouteExplorer to render multiple legs the way it already does
for Haystack's percentage-split routes.

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

Bumps [postcss](https://github.com/postcss/postcss) from 8.5.12 to 8.5.25.
- [Release notes](https://github.com/postcss/postcss/releases)
- [Changelog](https://github.com/postcss/postcss/blob/main/CHANGELOG.md)
- [Commits](https://github.com/postcss/postcss/compare/8.5.12...8.5.25)

---
updated-dependencies:
- dependency-name: postcss
  dependency-version: 8.5.25
  dependency-type: direct:development
...

Signed-off-by: dependabot[bot] <support@github.com>

AppClient error transformers matched app IDs with a substring check
(`"app=1142" in str(error)`), so a transformer for app 1142 could claim
errors for app 11423 and transform them with the wrong source map.

Match with a digit boundary (`app={id}(?!\d)`) instead.

Closes #296.