Decode tag values into buffer

[197g]

Two new methods for IfdDecoder:

• find_tag_buf(&mut self, &mut ValueBuffer) -> Result<Option<Entry>, _>
• find_tag_bytes(&mut self, buf: &mut [u8], offset: u64) -> Result<Option<usize>, _>

Both will only access bytes belonging to the tag entry if present. The former returns the entry itself as convenience. It must be an option for case the tag is missing, which we want to signal as success and Entry occurs naturally. Most of the information would be present on the buffer afterwards though. The byte version returns the number of bytes that have been written. You're expected to have called find_entry to allocate the buffer so the entry itself will likely not provide new information.

Note that find_tag_buf is not implemented in terms of the other since there is a potential optimization that delays allocating the buffer until we found that the reader actually has enough data present. (Similar to <dyn Read>::read_to_end). We always enforce the limit so this is pure optimization left for later.