[fix](doc) trim/field/substring: bidirectional en/zh sync

[boluor]

Summary

Careful bidirectional sync for three pages where EN and ZH had each grown distinct non-redundant examples. Each addition was verified end-to-end against Doris 4.1.1; expected outputs are pasted from the cluster, not transcribed.

trim.md

• EN ← ZH: add example 9 — UTF-8 string with a UTF-8 trim pattern. Demonstrates that trim('ṭṛì ḍḍumai+++', 'ṭṛì') strips only the leading ṭṛì (trailing +++ doesn't match the pattern so trim stops there). EN previously had only a default-space UTF-8 example.
• ZH ← EN: add examples 8/9/10 — no-match returns original, repeated-pattern strips until exhausted, and asymmetric removal (multi-char pattern from both ends). EN had these corner cases as 4/7/8; ZH didn't.
• ZH fix: trailing ;; (double semicolon) on example 7 → single ;.

field.md

• EN ← ZH: extend the class_test setup with a NULL row, then add two examples that exercise NULL handling under FIELD-based custom ordering — DESC variant (NULL ends up last because FIELD = 0) and NULLS FIRST (NULL placement independent of sort direction). Updated the existing ASC example's expected output to include the NULL row that the new setup row produces.
• ZH ← EN: add the simple SELECT FIELD(2, 3, 1, 2, 5) example that introduces the 1-based position semantics before the ORDER BY use case.

substring.md

• EN ← ZH: add example 12 — substring(NULL, 1, 3) (NULL passed to the function directly; the MID alias variant was already example 8).
• ZH ← EN: add example 12 — SUBSTR('Hello World', 7, 5) showing the SUBSTR alias.

Scope note

Two other P1-both-extra pages remain intentionally not in this PR:

• strright.md — EN and ZH are already at parity; the only diff is a cosmetic "syntax block lists the RIGHT alias twice" formatting difference. Not worth a doc change.
• array-count.md — EN and ZH each have substantially different example sets covering the same teaching points with different idioms (e.g. EN uses size(array_filter(...)) > 0 to show the "wrap inner higher-order in a scalar" pattern; ZH uses recursive array_count(... array_count(...) ...)). Both are valid; mechanically merging them risks a confusing doc.

🤖 Generated with Claude Code