fix(replication): let RemoteDefaultExports suppress remote default exports (#55)

[dmann000]

What

New-PfbFileSystemReplicaLink -RemoteDefaultExports was a [switch], so it could only ever send remote_default_exports=true. Omitting it sent nothing, letting the array apply its own default (create the exports). There was no way to send false.

Consequence, hit on every share during an active field migration: each replicated file system came up with default NFS/SMB exports on the remote array that had to be deleted by hand.

Fix

• -RemoteDefaultExports becomes [Nullable[bool]], matching the module's existing tri-state idiom (e.g. New-PfbFileSystem -SnapshotDirectoryEnabled, -Writable).
• The query param is sent only when the caller bound it, guarding on presence ($PSBoundParameters.ContainsKey) not truthiness, so $false reaches the wire:
  • omitted → array default (unchanged behavior)
  • $true → remote_default_exports=true
  • $false → remote_default_exports=false (previously impossible)

Tests

Adds Tests/New-PfbFileSystemReplicaLink.Tests.ps1 — the cmdlet had no test file. Asserts the wire value for all three states, the required selectors, and that the parameter is a nullable bool rather than a switch. Green on Pester 6.0.1 (6/6).

Breaking change

The bare -RemoteDefaultExports switch form now requires a value: use -RemoteDefaultExports $true. Worth a CHANGELOG "breaking" note when the release is cut.

Scope

This is the replica-link half of #55, which is verified and safe. The create-side half (suppressing the array_server default exports on New-PfbFileSystem) is intentionally not included here: neither default_exports nor remote_default_exports is modeled in the committed capability map, and the raw spec is not committed, so the exact create-side wire shape (empty array vs bool vs query key) can't be confirmed from repo artifacts. The field user is compiling written API notes with the exact observation; that half stays tracked in #55 to land once the wire format is confirmed, rather than guessed.

Refs #55.