fix: preserve language_presets locale code keys from case conversion

[donnyyung]

Problem: Ran into an issue where I was running elevenlabs agents pull, making some changes, and then running elevenlabs agents push. Because I had brazilian portuguese in the language presets, the initial pull looked like

        "language_presets": {
            "pt_br": {
                "overrides": {
                    "agent": {
                        "first_message": "Olá! Estou aqui para ajudá-lo a encontrar a ferramenta de software certa. O que você está tentando resolver?"
                    }
                },
                "first_message_translation": {
                    "source_hash": "{\"firstMessage\":\"Hi there! I'm here to help you find the right software tool. What are you solving for? \",\"language\":\"en\"}",
                    "text": "Olá! Estou aqui para ajudá-lo a encontrar a ferramenta de software certa. O que você está tentando resolver?"
                }
            },

And then on the push ran into an issue because pt_br became ptBr and I got this error from the api:

  "Value error, Preset languages must be one of en, zh, es, hi, pt, fr, de, ja, ar, ko,
  id, it, nl, tr, pl, ru, sv, tl, fil, ms, ro, uk, el, cs, da, fi, bg, hr, sk, ta, vi,
  no, hu, af, hy, as, az, be, bn, bs, ca, et, gl, ka, gu, ha, he, is, ga, jv, kn, kk,
  ky, lv, lt, lb, mk, ml, mr, ne, ps, fa, pa, sr, sd, sl, so, sw, te, th, ur, cy, pt-br
  but got ptBr"

I made this change on a local fork and it worked so wanted to contribute upstream.

[PennyroyalTea]

lgtm, thanks for contribution!