Allow closing '>' of multiline nested type arguments to align with the opener (#15171)

[edgarfgp]

Problem

A nested, multiline type-argument list fails to parse when the inner closing > is aligned with the column of the opening type name. Adding one extra space before the > avoids it.

Fixes #15171.

Before

type Terminal =
    abstract onKey:
        IEvent<
            Foo<
                Bar * int
            >          // aligned with 'Foo' → error FS0010: Incomplete structured construct ... in type arguments
        > with get, set

After

type Terminal =
    abstract onKey:
        IEvent<
            Foo<
                Bar * int
            >          // now accepted
        > with get, set

Cause

The closing > at the same column as the enclosing sequence block's start, on a later line, was treated as a block separator (OBLOCKSEP), which broke the type-argument parse. A closing > of a type application is a closing bracket like ) or ], so it now belongs in isSeqBlockElementContinuator alongside RPAREN/RBRACK/RBRACE. Scoped to GREATER true (the type-app close), so the > comparison operator is unaffected.

[github-actions]

❗ Release notes required

You can open this PR in browser to add release notes: open in github.dev

---

✅ Found changes and release notes in following paths:

Warning

No PR link found in some release notes, please consider adding it.

Change path	Release notes path	Description
src/Compiler	docs/release-notes/.FSharp.Compiler.Service/11.0.100.md	No current pull request URL (#20003) found, please consider adding it

[T-Gro]

🤖 AI review (@expert-reviewer): no significant issues found. Please verify independently.

Verified

• The new arm | GREATER true -> true in isSeqBlockElementContinuator is correctly scoped: GREATER true is only emitted for a type-application / typar-decl closing > (lex filter), while the comparison operator stays GREATER false (lex.fsl). So the > comparison operator is unaffected, as claimed.
• Consistent with existing precedent: TokenRExprParen (LexFilter.fs:196) already classifies GREATER true as a right-paren-like closing bracket alongside RPAREN/RBRACK/RBRACE.
• The continuator is only consulted when the token sits exactly at the seq-block offside column on a later line (LexFilter.fs:1925-1934). A new seq-block element never legitimately begins with a type-app-closing >, so no needed OBLOCKSEP is suppressed.
• Nested type-argument seq-blocks route through the �lse branch (isSeqBlockElementContinuator), not the isTypeCtxt branch (isTypeSeqBlockElementContinuator), which is why the fix covers the reported �bstract member case.
• Test coverage is appropriate: a ComponentTests typecheck-success case plus a SyntaxTree parse baseline.

Minor (non-blocking)

• Consider a sibling test for �al/record-field type annotations using the same multiline aligned-> shape, to lock in that they share the same lex-filter path.