serde_derive_internals: Unicode-aware camelCase first-char lowering

[harvinderRathore]

Problem

#[serde(rename_all = "camelCase")] could panic when a field or
variant name begins with a non-ASCII character. The cause was byte
slicing ([..1]) on a UTF-8 string before lowercasing the first
character. Examples include: İstanbul (U+0130) and other multi-byte
scalars.

Fix

Replace byte-slicing with Unicode-aware lowering of the first Unicode
scalar using char::to_lowercase(), then append the remainder. This
handles multi-scalar mappings (e.g., İ -> i + U+0307) and avoids
UTF-8 boundary panics. Updated both CamelCase arms in
apply_to_variant and apply_to_field.

Compatibility

ASCII behavior is unchanged. Only the first scalar is lowercased (same
camelCase strategy as before). The change is confined to derive-time
rename logic.

Tests

• ASCII regressions for fields/variants remain unchanged.
• Greek Σ -> σ.
• Expanding lowercase for İstanbul -> i\u{307}stanbul.
• Non-cased scripts (Chinese/Han and Hindi/Devanagari) remain stable.
• Mixed Devanagari + ASCII snake_case -> परियोजनाName.

All crate and workspace tests pass locally.

Closes: #2953