If you want bold, use bold. Real bold. Not mathematical symbols. Those characters are intended for use in mathematics only. They are not, semantically, text at all.
U+1D5DB βΉπβΊ \N{MATHEMATICAL SANS-SERIF BOLD CAPITAL H}
U+1D5F2 βΉπ²βΊ \N{MATHEMATICAL SANS-SERIF BOLD SMALL E}
U+1D5F9 βΉπΉβΊ \N{MATHEMATICAL SANS-SERIF BOLD SMALL L}
U+1D5FC βΉπΌβΊ \N{MATHEMATICAL SANS-SERIF BOLD SMALL O}
U+1D43B βΉπ»βΊ \N{MATHEMATICAL ITALIC CAPITAL H}
U+1D452 βΉπβΊ \N{MATHEMATICAL ITALIC SMALL E}
U+1D459 βΉπβΊ \N{MATHEMATICAL ITALIC SMALL L}
U+1D45C βΉπβΊ \N{MATHEMATICAL ITALIC SMALL O}
Assistive technologies such as screen readers do not interpret these characters as text, but as mathematical symbols, therefore these would be read aloud letter-by-letter as βmathematical bold capital Hβ or similar. For an example of what this sounds like in practice, see a Tweet from Kent C. Dodds which includes the audio produced by such a screen reader (VoiceOver) when faced with these characters.
The fact that π counts as two characters is a separate issue, and is to do with the way that Twitter processes Unicode characters (all Unicode characters, be they text or not). As far as I can tell from the Twitter docs, it should count as one character now, as they count characters (not bytes) in Unicode NFC. However, maybe it was different when you asked the question.