Fix SFTP client states dropping unsent bytes after partial channel send

[yosuke-wolfssl]

Description

Several SFTP client request states (read, write, open, mkdir, setstat, rename) advanced their state machine as soon as wolfSSH_SFTP_buffer_send() returned, without confirming the request bytes had actually left the client. Two cases were mishandled:

1. Clamped partial send — SendChannelData sends fewer bytes than requested (idx < sz); the old code freed/advanced past the buffer, corrupting the request stream.
2. Socket back-pressure — the buffer reports fully consumed (idx == sz) but bytes remain queued in the SSH output buffer (wolfSSH_OutputPending), returning WS_WANT_WRITE.

Additionally, wolfSSH_SFTP_Rename used a ret <= 0 guard that treated a WS_SUCCESS (0) flush-only resume as terminal, swallowing a subsequent server rejection and returning success.

Addressed by f_5574.

Changes

• Added a shared helper wolfSSH_SFTP_buffer_send_finish() that returns WS_WANT_WRITE unless both the buffer is fully consumed and nothing remains in the SSH output buffer. Wired it into all six affected states so each preserves its state/buffer and resumes correctly.
• Changed the rename send guard from ret <= 0 to ret < 0 so a flush-only resume falls through to read the server's response instead of returning premature success.
• New tests to exercise new code path with new test hooks.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated no new comments.

[wolfSSL-Fenrir-bot]

Fenrir Automated Review — PR #1008

Scan targets checked: none
Failed targets: wolfssh-bugs, wolfssh-src

⚠️ Review incomplete — one or more scan targets failed before findings could be produced. See the Fenrir PR review detail page for logs.