The Hebrew and Greek terms for 'messenger' do have this natural overlap, and can cause contention in translation. The Latin Vulgate was the first translation which tried to separate the word into divine and human, by transliterating the Greek term αγγελος to create the Latin angelus for divine messengers, and 'translating' it properly for human messengers, typically as either nuntius or legatus. These terms were then used throughout the biblical text, reading 'angelos' back into Old Testament texts.
Most English translations follow this tradition in some form, maintaining the separation in order to preserve the familiar 'angel' terminology which we have inherited in our cultures. It is likely that some translations follow the 'split' used by these early Latin traditions, and evidently others determine the translation for themselves on a case-by-case basis (hence the KJV's variance in translation of Job 33:23 "If there be a messenger with him, one among a thousand...", where the Vulgate used 'angelus'). The problem is that we then read inherited traditions of 'angels' into ancient texts which predated these traditions and their associated theologies.
Personally I find your overall question a lot easier than its associated details:
Q: How should we best translate these terms?
A: These terms should read 'messenger' in every instance
If we choose to translate these terms according to the Latin 'angelus / nuntius / legatus' split in order to get to English, we're injecting our own meaning into the text. Our motive is usually a good one, because we want to provide more clarity to the reader - but the underlying text does not provide this clarity. It could be argued that the underlying texts in many instances provoke exactly your question: 'is this a divine being or a human being?' In such cases, if we make a decision as a translator, we risk stealing natural questions and over-writing them with our own theologies.
However, some translators will want to preserve 'angel' in some texts for traditional reasons (perhaps they think the passage is best tied-in with our inherited understanding of angels), and so I would advise that this only be done in the few cases where there is no doubt to its spiritual origin. Yet this approach in my view is still lacking, as for unfamiliar readers it would imply that 'messenger' was necessarily non-spiritual.
Question restatement: Are there hermeneutic principles which help us discover whether there is an intended 'spiritual' or 'human' source implied in the Greek and Hebrew terms for 'messenger', in texts such as these?
Above I have argued that I would suggest to preserve 'messenger' in every instance, but what tools do we then have to determine the nature of these messengers? The words in question evidently have ranges of meaning, and the Hebrew term particularly is found in texts ranging across hundreds of years, so what indicators can we use to determine its range or intention in a given text?
Local descriptors: Local contextual cues help us determine the nature of many appearances of these figures. As stated in the question, if a 'messenger' kills 185,000 men in a night, this is a good indication they are more than just a human messenger! (2 Ki 19:35) Similarly other words are sometimes used to make references explicit, such as "the angel of [God's] presence" in Isaiah 63:9.
Consistency of usage: For a given author or text, we may establish norms of their usage of a term. For instance, in Zechariah's visions the 'messengers' are established as divine, and indeed one is synonymised with 'the angel of the Lord' (Zech 1:11, 12:8). The Hebrew phrase the angel of the Lord is a spiritual figure named more than fifty times in the Tanakh, and so this descriptor helps to make these references clear.
Genre: In most cases this would probably be point #1, but unfortunately genre will not often tell us very much about the intended human/divine aspects of messengers in the Bible. In the scriptures we find a mixture of divine figures in historical texts and human figures in figurative texts, so genre can only give us minimal information about the typical range of meaning we may expect this word to convey in a given text.
Revelation 3 is perhaps more obviously supernatural because it follows from Revelation 1, where the entire text is prefaced by the 'messenger' sent to John from God. This is used as justification for the text which follows, and so reads quite plainly as a divine revelation of some description. Therefore a consistent hermeneutic of 'angelos' in the following chapters makes sense (see also chapter 7 for straightforward supernatural usage).
In cases where these and other indicators clearly show the nature of a 'messenger', it may be valid to differentiate accordingly in translation. However, I would not apply this to the verses you have highlighted. In Job 33:23 and Acts 12:15, the 'messenger' is vague enough that I would be wary to translate them as 'angel' outright. At the most, it would make sense to translate these as 'angel' but add a footnote to suggest 'messenger' as an alternative.
'Peter's angel' has been examined at length in Acts - Who/What is Peter's Angel? The current 'accepted answer' to this question suggests that it was a contemporary belief that angels would take the form of people, and this is certainly plausible, especially considering that the slave girl made her judgement based on the 'sound' of the voice and not the words. But this would be cast into doubt as it's the assertion of her opponents who think she's somewhat mad, and so it's difficult to know if they were really concluding a supernatural occurrence, or think that it's just a messenger sent from Peter and she's become confused. So again, I'd reserve judgement on that passage.