[ipproyectosysoluciones]

Implements #200 (status: needs-review / enhancement).

Adds managed emoji font support and Unicode rendering configuration to the Vim trainer simulator (simulator.go) for Linux multi-byte UTF-8 input.

Changes:

• simulator.go: isWordChar byte→rune; add runeAtBytePos, isEmojiRune, findNextEmoji; update moveWord* for UTF-8
• simulator_unicode_test.go (PASS)
• installer/e2e/test_emoji_navigation.sh
• AGENTS.md reference + contribution spec doc (English title per user request)
• Fix obsidian.nvim config (completion.blink deprecated; picker.name → snacks.picker; workspace clean)

Build: go clean. Tests: PASS.
Label intended: type:feature (not present in upstream repo; suggest adding).

[ipproyectosysoluciones]

Both findings from the deepseek review are fixed in 90a0b9a:

• moveEndOfWord overshoot (blocking): the loop now advances while the next rune is a word char, so it stops on the last word rune instead of one past it; the trailing adjust block (whose condition was inverted) is gone. The review case (hello world from col 0 → col 4) plus the second-word hop, end-of-line words, and the emoji word case are pinned in new regression tests, and the pre-existing TestSimulateMotions_EndOfWord is green again (it was red on the previous head).
• E2E script: now uses a repo-relative path, asserts the helpers the feature actually ships, and exits non-zero on failure. The unconditional PASS was hiding that findNextEmoji never existed — the script was reporting "available: 0" for it.
• isWordCharForTextObj: now takes the source line plus byte offset and decodes the real rune, so text objects classify multi-byte letters and emoji by their actual code points.

go test ./internal/tui/trainer/ and go build ./... pass locally.