Add trivial le_to_u8() endianness routine

[jmarshall]

In writing a switch like this one in bcf_fmt_array():

htslib/vcf.c

Lines 2386 to 2392 in 8d91938

	switch (type) {
	case BCF_BT_INT8: BRANCH(int8_t, le_to_i8, v==bcf_int8_missing, v==bcf_int8_vector_end, kputw(v, s)); break;
	case BCF_BT_INT16: BRANCH(int16_t, le_to_i16, v==bcf_int16_missing, v==bcf_int16_vector_end, kputw(v, s)); break;
	case BCF_BT_INT32: BRANCH(int32_t, le_to_i32, v==bcf_int32_missing, v==bcf_int32_vector_end, kputw(v, s)); break;
	case BCF_BT_FLOAT: BRANCH(uint32_t, le_to_u32, v==bcf_float_missing, v==bcf_float_vector_end, kputd(le_to_float(p), s)); break;
	default: hts_log_error("Unexpected type %d", type); exit(1); break;
	}

 
but in which the output types were unsigned rather than signed as in this one, I found that there was no le_to_u8() function. One could in fact just write BRANCH(uint8_t, *, …) in many such cases, but that is pretty obscure for the reader.

This is even more trivial than le_to_i8() but can be useful for the regular treatment of unsigned types in these sorts of switches.