Avoid use hard coded size of partition in tts_stream (AUD-5226)

[yyjdelete]

Fixes #1094

[yyjdelete]

The code is the same as the one in esp-sr and esp-skainet, and should make people possible to use tts data >3M(esp_tts_voice_data_xiaoxin.dat) without change the code of adf(change partitions.csv is still required)
https://github.com/espressif/esp-sr/blob/cc806480c093f1bf866068a70cd6255da62e8275/test_apps/esp-tts/main/test_chinese_tts.c#L72-L76
https://github.com/espressif/esp-skainet/blob/5e03e16b329d966717c5fca1df806486a0ee88bc/examples/chinese_tts/main/main.c#L56-L62

[jason-mao]

Thanks for your contribution. It's merged on afeb5ff