feat: return outcome of set and recrypt.

[antondalgren]

This commits adds a boolean return value from the underlying MMKV library to the JS environment for the set and recrypt calls.

closes #583

[mrousavy]

Is there any case where a set could fail? Unless there's some critical memory error or something I feel like this is a bit unnecessary 🤔

[antondalgren]

The set could fail e.g if there is not enough disk space left.

Follwing is a set of method calls that will show this:
https://github.com/Tencent/MMKV/blob/272e5aa1fb57b4ee8cc0ff37e70b7885d99832a7/Core/MMKV_IO.cpp#L579

auto ret = appendDataWithKey(data, key, isDataHolder);

in appendDataWithKey
https://github.com/Tencent/MMKV/blob/272e5aa1fb57b4ee8cc0ff37e70b7885d99832a7/Core/MMKV_IO.cpp#L797

return doAppendDataWithKey(data, keyData, isDataHolder, static_cast<uint32_t>(keyData.length()));

in doAppendDataWithKey
https://github.com/Tencent/MMKV/blob/272e5aa1fb57b4ee8cc0ff37e70b7885d99832a7/Core/MMKV_IO.cpp#L744-L747

    bool hasEnoughSize = ensureMemorySize(size);
    if (!hasEnoughSize || !isFileValid()) {
        return make_pair(false, KeyValueHolder());
    }

in ensureMemorySize
https://github.com/Tencent/MMKV/blob/272e5aa1fb57b4ee8cc0ff37e70b7885d99832a7/Core/MMKV_IO.cpp#L382-L397

return expandAndWriteBack(newSize, std::move(preparedData))

inexpandAndWriteBack
https://github.com/Tencent/MMKV/blob/272e5aa1fb57b4ee8cc0ff37e70b7885d99832a7/Core/MMKV_IO.cpp#L401C21-L429

        // if we can't extend size, rollback to old state
        if (!m_file->truncate(fileSize)) {
            return false;
        }

        // check if we fail to make more space
        if (!isFileValid()) {
            MMKVWarning("[%s] file not valid", m_mmapID.c_str());
            return false;
        }

As recrypt also will end up in the expandAndWriteBack function, recrypt may also fail.

Without this boolean return value, users of this library may think that the value that was set is guaranteed to be persisted, but in fact we can see that if the file can't be expanded, the write will return false and rollback to the old state without notifying the caller.

[mrousavy]

Good point. Hm, I'll think about this and come back soon

Anyways, the PR is good & well implemented, thank you so much!