system: add toOwned for char/byte views

[ZoomRmc]

Implements RFC #558 by separating two currently conflated operations on openArray[char]:

• substr(openArray[char]) now behaves like the other substr overloads and performs substringing with clamping.
• the old "copy this whole view into a new string" behaviour is now provided explicitly as toOwned.

This also adds toOwned(openArray[byte]), covering the common byte-view to string-copy case.

Properly closes #14810

Rationale

Before this change, substr(openArray[char]) did not actually perform substringing. It was effectively a full-copy helper for an already-formed view, which made it inconsistent with the rest of the substr family, As noted in RFC #558:

• substr(string; first, last) selects a range and clamps indices
• substr(string; first = 0) selects a suffix
• substr(openArray[char]) simply copied the entire input view unchanged

This PR eliminates this semantic mismatch and makes substr(openArray[char]) consistent with the existing string overloads, giving the copy operation its own name.

Why toOwned

I chose toOwned over toString or just copy because this operation is not generic stringification or formatting. The name emphasizes the ownership transfer and data copying performed. toString would be less clear and would raise questions "why not $?".

toOwned makes the important properties explicit:

• the result is a fresh owned string
• the source data is copied
• this is first of all a data-ownership/materialization operation, not a presentation operation

Behaviour and backward compatibility

• substr(openArray[char]; first = 0; last = int.high) selects a range within a view and clamps first/last like the string overloads do.
• substr() with default arguments still produces the same result as toOwned(), but the APIs now communicate different intent:
  • toOwned: copy this exact view, in full
  • substr: select a range from this view, with clamping

Due to added default argument values substr[openArray[char]] remains fully backward-compatible.

Tests

Added dedicated test coverage for string-view behaviour in:

• tests/stdlib/tstringview_owned_substr.nim

New file preferred over tsystem_misc as the whole matrix check is required.

[ZoomRmc]

Errors all look IC/NIF-caused, don't think they're caused by this patch.