Reach connector configuration: production resource limit boundary between devnet and live network profiles

Reach maps a single smart-contract program to multiple blockchain backends through its connector abstraction. A deployment that succeeds against the local devnet connector can fail on a production connector such as Algorand or Ethereum because the local backend commonly relaxes gas, fee, and timeout constraints that live consensus networks enforce. The unresolved decision is whether connector-specific resource limits are inherited from a local profile or must be explicitly re-declared for each production network. Without a single enforced boundary, omitting a production connector flag may leave the deployment silently bound to local defaults that are invalid remotely.

Live networks introduce fee markets, nonce management, and block-time variability that the simulated devnet does not reproduce, so identical participant code can diverge without source changes. Because CLI flags and environment variables are version-sensitive, the exact inheritance behavior depends on the installed Reach toolchain.

Does the connector abstraction provide a documented boundary that forces production networks to reject local-only resource defaults, or must every production deployment explicitly declare its own gas, fee, and timeout settings? If explicit declaration is required, which configuration surface takes precedence when both a local profile and a production connector flag are present?