feat(HNT-2086): add crawl-common utility helpers

[mmiermans]

Goal

HNT-2086

Add two small crawl-common helpers that later services reuse: a registrable-domain resolver for the discovery same-domain filter, and temporary per-environment config fallbacks. Both are pure functions wired through the utils barrel.

See docs/ARCHITECTURE.md under review as #14 for how same-domain filtering fits the discovery flow.

Implementation decisions

Decision	Approach	Why
Resolve the registrable domain with tldts	getRegistrableDomain returns getDomain with allowPrivateDomains set true, backed by the Public Suffix List	Matches content-monorepo, which resolves the eTLD+1 with getDomain from tldts in servers/curated-corpus-api/src/shared/utils.ts. The Public Suffix List handles multi-part suffixes such as .co.uk that matter for non-US publishers, and allowPrivateDomains treats platform suffixes such as blogspot.com as the boundary so two publishers on one platform stay distinct domains.
Return undefined rather than throw on an unparseable URL or a bare IP	The helper coalesces the null result to undefined	The discovery same-domain filter is a soft edge check, so a URL it cannot parse should be filtered out, not crash the worker. content-monorepo's equivalent getRegistrableDomainFromUrl throws instead because its callers want the failure surfaced.
Ship deployed-defaults as an explicit throwaway module	It maps dev, stage, and prod to a Memorystore IP and a GCP project id, returns empty otherwise, and is wired only as the fallback used when the env var is unset	The mozcloud chart does not yet inject the non-secret PROJECT_ID and REDIS_HOST. These move to the hnt-config configMap once the matching webservices-infra change lands, at which point this module and its two config call sites are deleted.