It is not tradition, it is a pragmatic approach, for a number of
reasons.
First: IPA is NOT perfect, especially for languages (like Polish)
which contain affricates contrasting with consonantal clusters (like
c : ts, č : tš, ʒ : dz, ǯ : dž etc.). IPA is clearly an
English-biased standard, uncomfortable when analysing e.g. Slavic
languages (this is the main reason for which IPA has NOT a wide
application in Slavistics). Or Nostratic, where affricates are
single sounds, not clusters as IPA suggests. Tell Altaicists,
Uralicists etc. to use IPA...
Second: we have no idea what exactly was phonetical values of
reconstructed sounds. When writing "e", or "a", we do not have IPA
[e], [a] in mind! Even worse, IPA has not a symbol for central [a]!
Writing [a] we suggest it was front, and writing [e] we suggest it
was mid-close (and not mid-open). There are no reasons for such a
suggestion. So, there is also no reason for writing IPA symbols
instead of the ones widely spread in literature.
Using Latin-like transliteration (but still not IPA!) for Greek is
quite a different matter. According to this convention (you may call
it tradition, or even "tradition"), we write kh for the Greek letter
khi/khei which is simple more comfortable than writing following
unnecessary IPA standards like a special symbol for aspiration.
Writing "kh" is far more comfortable, and does not lead to
misunderstandings at the same time. Also writing þ instead of the
stylized Greek letter theta (used for IPA) is more comfortable: I
have thorn on my keyboard layouts (it is used in Icelandic) while I
have not the Greek alphabet, and the more I do not have IPA symbols!
BTW, sometimes IPA is very far from being accurate: English c/k
(like in "cat") is often aspirated and should not be spelt with just
[k] (in fact, in most cases it is pronounced like Greek khei and not
like kappa).
Then, the Indo-Europeanists convention is as good and as bad as IPA
is (and sometimes it is even more accurate). But wide spreading in
the literature works in favour of it, and against using IPA.
No seeking is involved: the
clear Germanic cognates make it
perfectly obvious that the word is from a Proto-Germanic
Class VI strong verb *hlaþaną.
>
Or better *xlaθanã using IPA
conventions. I disagree with the Indo-Europeanists'
"tradition" (I'd rather call it vice) of using non-IPA
symbols such as the "thorn" as well as the Greek
alphabet for transcribing Ancient Greek.