feat(memory): migrate EmbeddingProvider to KMP-compatible TextEmbeddingExecutor, add :memory-md-vector, and hybrid backend support

- Replaced `EmbeddingProvider` with cross-platform `TextEmbeddingExecutor` for native target compatibility.
- Introduced `:memory-md-vector` module combining vector-cache and `.md` file-based memory systems (`hybrid` backend).
- Updated `SiglipEmbeddingProvider` to use KMP `TextEmbeddingExtractor` and streamlined compatibility via `asExecutor`.
- Added hybrid memory backend to `standalone`, supporting `.md` reconciliation with vector-cache for semantic

gradle/libs.versions.toml

@@ -97,8 +97,13 @@ kotlinx-io-core = { module = "org.jetbrains.kotlinx:kotlinx-io-core", version.re
 jvector = { module = "io.github.jbellis:jvector", version.ref = "jvector" }
 
 # --- text-embedding-kmp (pw.binom.ai.embeddingtext) — on-device SigLIP2 эмбеддинг через ONNX. ---
-# Артефакты публикуются под именами `-jvm` (KMP convention для JVM-таргета).
-text-embedding-api = { module = "pw.binom.ai.embeddingtext:api-jvm", version.ref = "text-embedding-kmp" }
+# `api` — KMP с jvm + android + linuxX64/Arm64 + macos + ios + mingwX64
+# (с 2026-09-21, когда мы добавили нативные цели в text-embedding-kmp:api).
+# Используется из :memory-md-vector и :memory-vector напрямую через
+# `libs.text.embedding.api` (без суффикса `-jvm` — Gradle сам выберет
+# нужный variant под target).
+# `siglip` — JVM+Android only (onnx-runtime), подключается в jvmMain.
+text-embedding-api = { module = "pw.binom.ai.embeddingtext:api", version.ref = "text-embedding-kmp" }
 text-embedding-siglip = { module = "pw.binom.ai.embeddingtext:siglip-jvm", version.ref = "text-embedding-kmp" }
 
 # --- Логирование: kotlin-logging (тонкая обёртка над slf4j-api) + logback-classic (binding). ---