Parser: capture named fields block separators

[edgarfgp]

Description

In this PR I propose we captured the block separators in the same way we do it when we do it records

This is a pre-requisite for #18833 where we need to able to track the separator.

BEFORE

• No block separator captured

// Union cases: Simple Ident
SynArgPats.NamePatPairs(
    pats = [(Ident("a"), Option<Range>, SynPat)]
)

// Records: Complex tuple with LongIdent
SynPat.Record(
    fieldPats = [((LongIdent, Ident("X")), Option<Range>, SynPat)]
)

AFTER

• Block separator captured

// Unified representation across all constructs
SynArgPats.NamePatPairs(
    pats = [NamePatPairField]  // Consistent type
)

SynPat.Record(
    fieldPats = [NamePatPairField]  // Same type used
)

// Both use the same field name type
NamePatPairField(
    fieldName = SynLongIdent([a], [], [None])  // Unified representation
)

// Union case with named fields - now uses SynLongIdent and captures separators
match x with
| A(a = a; b = b; c = c) -> 2
//  ^SynLongIdent for field names, separators captured

// Record pattern - unified representation
let { X = x; Y = y } = record
//    ^Same SynLongIdent structure

// Future possibility: qualified field names
// | SomeCase(Module.field = value) -> ...  // Now structurally possible

Checklist

• Test cases added
• Release notes entry updated

[edgarfgp]

/run ilverify

[github-actions]

🔧 CLI Command Report

• Command: /run ilverify
• Outcome: success

✅ Command succeeded, no changes needed.