fix(docker): install OpenTofu >= 1.10 instead of alpine's 1.7.2

[sebasnallar]

Problem

The lambda scope inits its S3 backend with use_lockfile=true (lambda/scope/tofu/provider/aws/setup), which requires OpenTofu ≥ 1.10. The previous Dockerfile installed opentofu from alpine 3.20's apk, which is 1.7.2 — so create-scope would fail at tofu init with an unsupported-argument error. This regressed the mise-based test image (.mise.toml → opentofu = latest, which was 1.10.x).

Fix

Drop opentofu from the apk line; pull the official static tofu binary instead, pinned to 1.10.6, per build arch (TARGETARCH). Verified locally:

tofu:     OpenTofu v1.10.6
aws:      aws-cli/2.15.57
gomplate: gomplate 3.11.7
worker:   /usr/local/bin/worker

Follow-up

The TF package pin (implementation-aws/.../scope_definition.tf) must point at the digest of the image built from this fix — not the current v0.3.0 (tofu 1.7.2). It'll get a new digest once this merges + releases (or a manual rebuild).