Spec compliance for in-memory representation of span and trace IDs

[fbogsany]

The specification for SpanContext requires that:

TraceId A valid trace identifier is a 16-byte array with at least one non-zero byte.

SpanId A valid span identifier is an 8-byte array with at least one non-zero byte.

Our representation is 16 and 32 character hex strings:

opentelemetry-ruby/api/lib/opentelemetry/trace.rb

Lines 11 to 42 in 79b047a

	module Trace
	# An invalid trace identifier, a 16-byte array with all zero bytes, encoded
	# as a hexadecimal string.
	INVALID_TRACE_ID = ('0' * 32).freeze
	# An invalid span identifier, an 8-byte array with all zero bytes, encoded
	# as a hexadecimal string.
	INVALID_SPAN_ID = ('0' * 16).freeze
	# Generates a valid trace identifier, a 16-byte array with at least one
	# non-zero byte, encoded as a hexadecimal string.
	#
	# @return [String] a hexadecimal string encoding of a valid trace ID.
	def self.generate_trace_id
	loop do
	id = Random::DEFAULT.bytes(16).unpack1('H*')
	return id unless id == INVALID_TRACE_ID
	end
	end
	# Generates a valid span identifier, an 8-byte array with at least one
	# non-zero byte, encoded as a hexadecimal string.
	#
	# @return [String] a hexadecimal string encoding of a valid span ID.
	def self.generate_span_id
	loop do
	id = Random::DEFAULT.bytes(8).unpack1('H*')
	return id unless id == INVALID_SPAN_ID
	end
	end
	end
	end

We should conform to the spec. This actually makes a few things better, including fewer allocations overall and making translation to things that don't use hex strings much easier to understand and prove correct.