What makes a font a legacy font
Unicode gives every letter of every Indian script its own number, so ఆ or अ is the same character whatever font draws it. A legacy font does not use those numbers. It borrows the numbers of English letters and punctuation and draws its own shapes there, so a document set in it is only readable while that exact font is drawing it.
These fonts were the standard for Indian-language DTP for a long time, so newspapers, government forms, books and court papers are full of them. The ones met most often are Anu, Priyaanka and Kranthi for Telugu; Kruti Dev, DevLys and Chanakya for Hindi and Marathi; SutonnyMJ, the Bijoy font, for Bengali; TAB-ELCOT-Kovai and Bamini for Tamil; Nudi for Kannada; LMG Arun for Gujarati; and Shree-Lipi across several scripts.
Three ways to check by hand
Copy a line and paste it into Notepad. Latin letters and symbols where the page shows an Indian script is the clearest sign there is. Search the PDF for a word you can see on the page: a legacy PDF will not find it, because the word is not in the file as text. Or open the font list, which Adobe Acrobat Reader shows under File › Properties › Fonts and the pdffonts command prints on the command line, and look for a name such as Kruti Dev 010 or Priyaanka.
The font list is the weakest of the three. Fonts are often embedded under names that mean nothing: one 154-page Telugu novel carries Anu fonts named TT1D00O00 and TTE2371118O00.
What the analyser checks instead
It reads the codes each font draws against the conversion tables IndicPDF has, for Anu, Kruti Dev, Akruti Marathi, Shree-Lipi, Bijoy, TAB-ELCOT-Kovai, TSCII, Bamini, Nudi, LMG Arun, ML-TT Karthika, Akruti Oriya and Asees, and names the family when the codes read as that script. Where the codes cannot be read but the font name matches a table, it accepts the name only if the font’s own text is not already Unicode Indian script, so a Unicode document is not flagged because of what its font is called. DevLys and Chanakya are read too, DevLys as a build of Kruti Dev. It also recognises Shivaji, Akruti (other than Akruti Oriya and Akruti Marathi) and Shree-Lipi by name, and says plainly that there is no conversion table for them yet.
What it cannot see
A legacy font none of the tables has met, embedded under a meaningless name, can pass without being named. If the analyser names no legacy font but its scripts panel says the text is almost all Latin while you can see Telugu, Hindi or Tamil on the page, trust your eyes: that combination is a legacy encoding. Nor can a font say which language a Devanagari document is in, because Kruti Dev is used for Hindi, Marathi and Sanskrit alike.
Why check before converting
The wrong tool fails quietly. A general-purpose PDF to Word converter trusts the file’s map, writes the same symbols into Word and reports success; nobody notices until someone tries to edit or search the result. A legacy PDF needs Legacy PDF to DOCX, a scan needs OCR, and a locked PDF needs its owner’s permission, and the analyser says which of those you have in one pass.